1

A Comprehensive Overview on 5G-and-Beyond Networks with UAVs: From Communications to Sensing and Intelligence Qingqing Wu, Jie Xu, Yong Zeng, Derrick Wing Kwan Ng, Naofal Al-Dhahir, Robert Schober, and A. Lee Swindlehurst

Abstract—Due to the advancements in cellular technologies intelligence is also essential for 5G-and-beyond 3D heterogeneous and the dense deployment of cellular infrastructure, integrating wireless networks with coexisting aerial and ground users. In unmanned aerial vehicles (UAVs) into the fifth-generation (5G) this paper, we provide a comprehensive overview of the latest and beyond cellular networks is a promising solution to achieve research efforts on integrating UAVs into cellular networks, safe UAV operation as well as enabling diversified applications with an emphasis on how to exploit advanced techniques (e.g., with mission-specific payload data delivery. In particular, 5G intelligent reflecting surface, short packet transmission, energy networks need to support three typical usage scenarios, namely, harvesting, joint communication and radar sensing, and edge enhanced mobile broadband (eMBB), ultra-reliable low-latency intelligence) to meet the diversified service requirements of next- communications (URLLC), and massive machine-type communi- generation wireless systems. Moreover, we highlight important cations (mMTC). On the one hand, UAVs can be leveraged as directions for further investigation in future work. cost-effective aerial platforms to provide ground users with en- hanced communication services by exploiting their high cruising Index Terms—Unmanned aerial vehicle (UAV), 5G-and-beyond altitude and controllable maneuverability in three-dimensional 3D cellular networks, aerial-terrestrial integration, communica- (3D) space. On the other hand, providing such communication tion, sensing, network intelligence. services simultaneously for both UAV and ground users poses new challenges due to the need for ubiquitous 3D signal coverage as I.INTRODUCTION well as the strong air-ground network interference. Besides the requirement of high-performance wireless communications, the The global market for commercial unmanned aerial vehicles ability to support effective and efficient sensing as well as network (UAVs), also known as drones, has grown significantly over the last decade and is projected to skyrocket to 45.8 billion Q. Wu is with the State Key Laboratory of Internet of Things for Smart dollars in 2025 from 19.3 billion dollars in 2020 [2]. The City and Department of Electrical and Computer Engineering, University of major driving factors behind such a dramatic market size Macau, Macau, China 999078 (email: [email protected]). J. Xu is with the Future Network of Intelligence Institute (FNii) and the growth are the steadily decreasing manufacturing costs and School of Science and Engineering, The Chinese University of Hong Kong the increasing number of applications in a broad range of (Shenzhen), Shenzhen 518172, China (e-mail: [email protected]). civilian and commercial sectors, including surveillance and Y. Zeng is with the National Mobile Communications Research Laboratory, Southeast University, Nanjing 210096, and also with the Purple Mountain monitoring, aerial imaging, precision agriculture, smart lo- Laboratories, Nanjing 211111, China (e-mail: yong [email protected]). gistics, law enforcement, disaster response, and prehospital D. W. K. Ng is with the School of Electrical Engineering and Telecommu- emergency care. Particularly, as announced by the Federal nications, the University of New South Wales, NSW 2052, Australia (e-mail: [email protected]). Aviation Administration (FAA) during its press conference N. Al-Dhahir is with the Department of Electrical and Computer Engi- on the “Drone Integration Pilot Program” in Washington on neering, the University of Texas at Dallas, TX 75083-0688, USA (e-mail: November 8, 2017 [3]: “The need for us to integrate unmanned [email protected]). R. Schober is with the Institute for Digital Communications, Friedrich- aircraft into the National Airspace System (NAS) continues to arXiv:2010.09317v2 [cs.IT] 12 Jun 2021 Alexander-University Erlangen-Nurnberg, 91058 Erlangen, Germany (e-mail: be a national priority. After the hurricanes, drones became [email protected]). a literal lifeline. They gave us an operational window that A. L. Swindlehurst is with the Center for Pervasive Communications and Computing, University of California, Irvine, CA 92697, USA (e-mail: was a game changer at every level.” This national program [email protected]). was launched to further explore the expanded use of drones, The work of Q. Wu was supported in part by the Macau Science including beyond-visual-line-of-sight (BVLOS) flights, night- and Technology Development Fund, Macau SAR, under SKL-IOTSC-2021- 2023, 0119/2020/A3, 0108/2020/A, and the Guangdong NSF under Grant time operation, and flights over people. Later, during the Notre 2021A1515011900. The work of J. Xu was supported by the National Natural Dame Cathedral fire in 2019 [4], two UAVs equipped with Science Foundation of China under grants 61871137 and U2001208, and high-resolution thermal imaging cameras were dispatched in the Science and Technology Program of Guangdong Province under grant 2021A0505030002. The work of Y. Zeng was supported by the Natural BVLOS environments to help firefighters gauge the scene Science Foundation of China under Grant 62071114, by the Fundamental through the billowing smoke and effectively position firehoses Research Funds for the Central Universities of China under grant number to combat the blaze in real-time. Very recently, UAVs have 3204002004A2, and also by the “Program for Innovative Talents and En- trepreneur in Jiangsu” under grant number 1104000402. The work of D. also been applied worldwide to combat the spread of COVID- W. K. Ng was supported by funding from the UNSW Digital Grid Futures 19, e.g., to facilitate communication/broadcast information, Institute, UNSW, Sydney, under a cross-disciplinary fund scheme and by the disinfect outbreak-affected areas, deliver critical supplies, and Australian Research Council’s Discovery Project (DP210102169). The work of A. L. Swindlehurst was supported by U.S. National Science Foundation measure body temperatures [5]. In December 28, 2020, the grant ECCS-2030029. FAA further released two long-awaited drone rules. One is 2

Year 2000 2005 2010 2015 2020 2025

Radio Frequency (uplink only) Manual controlled drone via RF (non-Wi-Fi)

Wi-Fi (up- & downlink) Drone control incl. video downlink (local Wi-Fi network)

4G/LTE Mbit/s 10.000 Same as Wi-Fi-operated but via 10.000 cellular network 4G/LTE

8.000 5G Same as 4G but via cellular network 5G 6.000

4.000 Up to 100 times higher bitrate at five times lower latency, compared to regular 4G/LTE 2.000 1.300 1.000 600 100 0 Max. Download Speed per Technology Wi-Fi Wi-Fi 4G/LTE 4G/LTE 5G 2.4GHz 5.2GHz Advanced

Fig. 1. The evolution of drone connectivity and the role of 5G [1]. on remote identification for UAVs and the other is on UAV hand, to guarantee safe and efficient flight operations of mul- operations at night and over people. These new rules are tiple UAVs, it is of paramount importance to provide secure expected to address the safety, security and privacy concerns and ultra-reliable communication links between the UAVs and while advancing opportunities for innovation and utilization of their ground pilots or control stations for conveying command UAV technology. and control signals, especially in BVLOS scenarios. Moreover, In practice, depending on the size, weight, wing configura- some practical UAV applications (e.g., real-time drone video tion, flying duration, etc., UAVs can be classified into different filming and streaming to ground entities) require very high categories, such as large UAV versus small/mini UAVs, fixed- data rates in the air-to-ground payload communication links. wing versus rotary-wing UAVs, etc. [6]. Each type of UAVs Fortunately, these requirements can in principle be largely generally possesses a set of unique characteristics and thus met by cellular networks, thanks to their densely deployed may be suitable for different application scenarios. For ex- communication infrastructure as well as advanced 5G-and- ample, fixed-wing UAVs have higher maximum flying speed, beyond technologies. On the other hand, because of advances greater payloads, and longer flying endurance than rotary-wing in communication equipment miniaturization as well as UAV UAVs, whereas the former requires a runway or launcher for manufacturing, mounting compact and lightweight base sta- takeoff/landing and it is also difficult for them to hover at tions (BSs) or relays on UAVs becomes increasingly feasible. a fixed position. In contrast, rotary-wing UAVs not only can This leads to new types of flying aerial platforms that can be take off/land vertically but also remain static at desired hover- exploited to improve the quality of services for terrestrial users ing locations, which renders them appealing for applications as well as to satisfy the need for on-demand deployment to such as monitoring. A detailed overview on different UAV address, e.g., temporary or unexpected events. This has led to classifications and applications was provided in [7]. two promising research paradigms for UAV communications, namely, UAV-assisted cellular communications and cellular- Wireless communication is an essential technology to un- connected UAVs [8], where UAVs are integrated into cellular lock the full potential of UAVs in numerous applications and networks as aerial communication platforms and aerial users, has thus received unprecedented attention recently [6], [8]– respectively. As such, integrating UAVs into cellular networks [10]. As shown in Fig. 1, UAV communication technologies is believed to be a win-win technology for both UAV-related have evolved from the early direct link to the most recent 5G industries and cellular network operators, which not only technologies in the course of the past twenty years, with signif- creates plenty of new business opportunities but also benefits icantly enhanced communication rate. Although technologies the communication performance of three-dimensional (3D) such as direct link, WiFi, and satellite communications are wireless networks. still useful in some remote scenarios where cellular services are unavailable, it is believed that exploiting the thriving 5G- and-beyond cellular networks to support UAV communications is the most promising and cost-effective approach, especially when the number of UAVs grows dramatically. On the one 3

A. Communication, Sensing, and Intelligence in 3D UAV a cost-effective manner [19]–[22]. For example, by leveraging Networks mobility to increase the operating altitude or bypass obstacles, According to International Mobile Telecommunications- UAV-BSs/users are able to intelligently reposition themselves 2020 (IMT-2020), 5G networks are expected to support a to avoid signal blockage to ground nodes caused by e.g., diverse variety of use cases in three main pillar categories high-rise buildings. This is practically appealing for millimeter [11]: wave (mmWave) communications that are expected to pro- vide eMBB services. Furthermore, for IoT/sensor networks, a • Enhanced mobile broadband (eMBB) which aims to hybrid network architecture composed of high altitude UAV- support data-intensive use cases such as virtual and aug- BSs/relays and their ground counterparts is able to achieve mented reality (V/AR), requiring high data rates across a wider ground coverage. In addition, UAV-enabled mobile data wide coverage area [12]. collectors can move close to IoT devices to collect data, such • Ultra-reliable and low-latency communications (URLLC) that their transmit energy is minimized. This is particularly which aims to support mission-critical applications such beneficial in mMTC scenarios where connectivity and energy as remote surgery, autonomous vehicles, and the Tactile consumption are critical factors [23]. Moreover, non-trivial Internet, requiring both ultra-high reliability and short tradeoffs may exist in leveraging these advantages [20], [24]– delay [13], [14]. [26]. For instance, although exploiting the high mobility of • Massive machine-type communications (mMTC) which a UAV access point/data sink via trajectory design could im- aims to provide connectivity for massive numbers of prove the communication throughput, this comes at the cost of power-limited devices with heterogenous traffic profiles exceedingly long user access delays [24], [26], which thus may such as in the industrial Internet-of-Things (IoT) [15]. not be appropriate for delay-sensitive URLLC applications. Although some of the above scenarios have been well Furthermore, to reap the promised benefits of UAV-assisted studied for terrestrial wireless networks [11], their techniques cellular communications, there are still many important prob- and results may not be directly applicable to future 3D wireless lems that need to be addressed, including more accurate air- networks featuring both UAVs (either as aerial users or as ground channel modelling, traffic-adaptive UAV deployment communication platforms) and ground users, due to their sig- and/or trajectory design, energy consumption modelling and nificantly different operating environments. More importantly, energy-efficient design, and high-speed and reliable backhaul- the new degrees of freedom they introduce for system design ing. have been unexplored previously but can help further enhance Besides the requirement for high-performance wireless the communication performance. communications, the ability to support effective and efficient On the one hand, providing these services simultaneously sensing is also essential for realizing the vision of integrating for both UAVs and ground users imposes new technical chal- UAVs into 5G-and-beyond networks. On the one hand, sensing lenges [8], [16]–[18]. For example, since UAVs usually operate techniques are helpful in achieving safe UAV operation and at much higher altitudes than conventional terrestrial users, intelligent air traffic management. On the other hand, UAVs exploiting the high beamforming gain via multiple antennas can be leveraged as aerial sensing platforms to collect infor- to provide eMBB service to UAVs requires ground BSs to mation in the sky. Recently, radio-based sensing technologies have the capability of steering the beam not only in azimuth have received increasing attention due to their ability to but also in the elevation plane. As such, the typical downtilted achieve contactless and privacy-preserving object detection BS antennas need to be reconfigured, since currently they cater and environmental monitoring. Although the principles and only to ground coverage as well as suppression of inter-cell objectives of radio-based sensing and radio-based communica- interference in long-term evolution (LTE) systems. In addition, tion systems are different, there have been significant research the line-of-sight (LoS) dominated air-ground channels inherent efforts to investigate their coexistence, cooperation, and joint to UAVs at high altitudes lead to smaller path loss due to design, thus leading to a new paradigm referred to as joint less severe shadowing and multi-path fading. As a result, communication and sensing (JCAS) [47]–[50]. This is because leveraging spectrum sharing to simultaneously support both the same wireless infrastructure, RF hardware, and spectrum UAVs and ground users in mMTC scenarios will generate can be shared by both systems, which avoids the high costs of significant ground-air interference to UAVs in the downlink, building dedicated wide-area sensing infrastructure and also while in the uplink, UAVs will cause stronger interference to helps unleash the maximum potential of cellular networks. a large number of ground users even if they are distributed Despite these appealing advantages, the research on JCAS in in the network [17]. Moreover, high UAV mobility generally UAV communication networks is still in an early stage and results in more frequent handovers and time-varying wireless there are many interesting and important problems that are backhaul links between UAVs and ground BSs/users, which open and require careful investigation. poses practical challenges in URLLC scenarios. As such, it Recently, artificial intelligence (AI) has been considered as is imperative to develop new techniques and/or new designs, another key enabling technology for 5G-and-beyond wireless e.g., efficient handover management and resource allocation, networks integrated with UAVs [51], [52]. This is mainly to tackle the above issues to guarantee seamless service due to its great potential to efficiently address challenging provisioning in the 3D space. problems involving large amounts of data in system design and On the other hand, the emerging 3D wireless networks also optimization, which positions AI-based approaches as pow- introduce new design opportunities to better serve users in erful tools to facilitate highly dynamic UAV communication 4

TABLE I LISTOFMAININDUSTRYPROGRESS, PROTOTYPES, AND PROJECTS RELATED TO UAV COMMUNICATION NETWORKS.

Company Year Main activity and achievement Qualcomm 2016 Feasibility proof of drone operation over commercial LTE networks at up to 400 feet (122 m) [27]. Intel and AT&T 2016 Demonstrate the world’s first LTE-connected drone at the 2016 Mobile World Congress [28]. Ericsson and China Mobile 2016 Claim the worlds first 5G-enabled drone prototype field trial in WuXi of China [29]. Nokia 2016 Design the flying-cell (F-Cell) which enables highly efficient “drop and forget” small cell deployments [30]. Verizon 2017 Conduct flight tests using a “flying cell site” aboard a drone to supply an LTE network if severe weather knocks out more traditional cellular network infrastructure [31]. Facebook 2017 Test “Tether-Tenna” to beam down the Internet from helicopters [32]. Huawei 2019 Announce 5G SkySite that contains a 5G remote radio unit (RRU) to provide signal coverage to 20-30 km2 area while flying at 100 metres [33]. Research project Start year Main objective PercEvite 2017 Develop a lightweight and energy-efficient sensor, communication, and processing suite for small drones for autonomously detecting and avoiding “ground-based” obstacles and flying objects [34], [35]. DroC2om 2017 Design and evaluate an integrated cellular-satellite system architecture for datalinks so that command and control information can be reliably transferred in support of functions and specific procedures for airspace access [36], [37]. SECOPS 2017 Define an integrated security concept for drone operations that ensures that security risks are mitigated to an acceptable level [38]. ABSOLUTE 2017 Leverage aerial BSs along with terrestrial and satellite communications to enhance the ground network capacity, especially for public safety in emergency situations [39], [40]. DARE 2017 Conduct advanced research on a new distributed autonomous and resilient emergency management system based on wireless sensor, ad-hoc, and cellular networks [41], [42]. 5G!Drones 2019 Test use cases for vertical industry applications (IoT, industry 4.0, autonomous cars, etc.) on 5G test platforms [43], [44]. AERPAW 2020 Build an aerial wireless experimentation platform spanning 5G technologies and beyond and enable cutting-edge research with the potential to create transformative wireless advances for aerial systems [45], [46]. networks. For example, traditional off-line and model-driven as the number of UAVs in the network grows. Nokia’s flying- trajectory design approaches usually require accurate and cell (F-Cell) is an experimental small cell that does not need tractable communication models with perfect global knowl- any physical wires. In particular, a UAV obtains power from edge of all system parameters. This limits their applicability the surface-mounted solar panels and communicates with the in practical scenarios with open operating environments and carrier’s core network over a high-speed wireless link, thus time- and spatial-varying traffic, especially when considering overcoming the challenges in backhaul cabling, deployment, the real-time movement of UAVs and ground users in 3D and expenses. While Qualcomm, Intel, AT&T, Nokia, and space. Instead, by leveraging, e.g., deep reinforcement learn- Verizon used existing 4G/LTE cellular networks, Ericsson, ing, UAVs can be endowed with the capability of predicting China Mobile, and Huawei have conducted UAV experiments future network states in an online manner and thus adapt by leveraging more advanced 5G technologies. In particular, the communication resource allocation as well as the UAV Huawei designed a 5G BS in 2019, called SkySite, which is trajectories based on the network dynamics. Furthermore, a UAV mounted remote radio unit. Different from Nokia’s F- UAVs with AI-embedded systems are also helpful in emerging Cell, Huawei’s SkySite has a tethering cable and it is used to applications such as edge computing, where multiple UAVs provide a stable power supply as well as ultra-high speed and could work collaboratively either as aerial edge servers or edge secured backhaul from the ground, thus enabling unlimited devices to achieve efficient data/computation offloading. endurance with high communication performance. There have also been several pilot projects launched in B. Industry Progress, Projects, and Standardization recent years to advance UAV research and UAV field tests, as Recently, UAV communications have also drawn significant also shown in Table I. In particular, in the ABSOLUTE project, attention from industry, as shown in Table I. Qualcomm’s field UAVs are leveraged as aerial BSs for terrestrial and satel- test was followed by a trial report released in May 2017. The lite communications to enhance the ground network capacity main focus of the report are the evaluation of the downlink and extend the signal coverage, especially for public safety signal-to-interference-plus-noise ratio (SINR) distribution and applications in emergency situations. Besides, the 5G!Drones the study of the impact of power control and resource parti- project was funded by the European Commission in 2019 to tioning enhancements on uplink interference and throughput drive UAV vertical application trials by leveraging advanced 5

TABLE II LISTOFMAIN 3GPP STANDARDIZATION PROGRESS ON UAV COMMUNICATION NETWORKS [53]–[55].

3GPP Release No. Study/Working Item Release 15, TS 36.331 Enhanced LTE Support for Aerial Vehicles Release 16, TS 22.125 Remote Identification of Unmanned Aerial Systems Release 17, TS 22.125 5G Enhancement for UAVs Release 17, TR 23.755 Study on Application Layer Support for Unmanned Aerial System (UAS) Release 17, TR 23.754 Study on Supporting Unmanned Aerial Systems Connectivity, Identification, and Tracking

5G features, and involves 20 industrial and academic partners and AI technologies are considered to further enhance the from 8 European countries, covering verticals, commercial capabilities of future UAV cellular networks, but were not network operators, networking industry, research centers, and covered in previous overview and survey papers. To facilitate universities. Specifically, this project exploits network slicing and inspire future research, various promising directions for as the key component to simultaneously run the three types of further investigation are also highlighted. UAV services (eMBB, URLLC, and mMTC) on the same 5G Next, we provide an overview on the aforementioned five infrastructure, demonstrating that each UAV application runs research areas respectively from Section II to Section VI. independently and does not affect the performance of other Specifically, in Section II, we highlight how emerging tech- UAV applications, while providing different 5G services. nologies, such as intelligent reflecting surface (IRS), can be Besides, the third-generation partnership project (3GPP) leveraged to support eMBB applications in 3D space and has made significant efforts to ensure that cellular networks also discuss the potential of exploiting massive multiple- will meet the connectivity demands of UAVs [53]–[55], as input multiple-output (M-MIMO) and mmWave. In Section shown in Table II. In 2017, 3GPP approved a study item on III, we provide an overview on short-packet communications enhanced LTE support for UAVs, where the main objective (SPC) to guarantee URLLC in UAV networks. In Section IV, was to identify key challenges for using current LTE networks we consider non-orthogonal multiple access (NOMA), energy with downtilted BS antennas to provide connectivity to UAVs. harvesting, and energy efficient designs for the support of the To further meet the requirements of connectivity, identifica- massive connectivity. In Section V, we focus on radio-based tion, and tracking of UAVs, 3GPP recently considered the sensing in wireless networks with UAVs and introduce a joint application of 5G networks in Release 17 [55]. Several other UAV communication and sensing approach. In Section VI, standardization bodies and working groups have also devoted we review machine learning methods particularly for UAV substantial efforts to develop different UAV specifications trajectory and communications design, discuss the computa- [6], including the International Telecommunication Union tion offloading design for UAVs with mobile edge computing Telecommunication (ITU-T) standardization sector, the Euro- (MEC), and consider UAV-based distributed edge machine pean Telecommunications Standards Institute (ETSI), and the learning. In each section, directions for future research are IEEE Drones Working Group (DWG). Furthermore, to foster also provided. Finally, the paper is concluded in Section VII. the research and innovation surrounding the study, design, and development of aerial communications, IEEE Vehicular Tech- nology Society (VTS) created an ad hoc committee on drones II.ENHANCED MOBILE BROADBAND and IEEE Communication Society (ComSoc) established an eMBB is a natural evolution of 4G/LTE networks and emerging technology initiative focusing on aerial users and will provide higher data rates and system capacity than cur- networks [56], [57]. One of the target applications of the rent mobile broadband services, while guaranteeing moderate initiative is public safety, i.e., using aerial communications reliability, e.g., packet error rates (PERs) on the order of to deliver additional cellular coverage, for example during 10−3 [11]. In particular, 5G cellular networks are expected emergency situations, or to help first responders by providing to support a peak data rate of 10 Gbits/s for eMBB appli- advanced services. Moreover, two standards working groups cations, which in principle is sufficient for supporting high- jointly sponsored by IEEE ComSoc and IEEE VTS focus on rate UAV communication applications such as real-time video aerial communications and networking standards [58]. streaming and data relaying. On the other hand, UAV-enabled aerial platforms provide new degrees of freedom to further C. Objectives and Contributions enhance the communication throughput or satisfy throughput It is worth noting that a number of papers in the literature requirements in a cost-effective manner. For terrestrial cellular provide overviews or surveys on UAV communication research networks, a variety of viable solutions have been proposed to [6], [8], [9], [16], [19], [54], [59]–[72]. Unlike these works, meet the stringent eMBB service requirements [73], namely, this paper explicitly focuses on identifying the new major enhance spectral efficiency by deploying massive active trans- challenges posed by the eMBB, URLLC, and mMTC use mit/receive antennas and/or passive reflecting elements such cases to UAV communication in 3D space, and highlights the as M-MIMO and IRS; exploit unused new spectrum such most promising solutions to tackle them. Moreover, sensing as mmWave communications; reduce the transmitter-receiver 6 distance and improve frequency reuse such as ultra-dense net- Another promising approach to support rate-demanding works (UDNs) and device-to-device (D2D) communications eMBB services in 3D space is to leverage the enormous chunks [74]–[76]. In this section, we focus on the exploitation of first of new spectrum available in the mmWave bands [89], [90]. two paradigms in 3D UAV cellular networks, with a particular Several inherent limitations of mmWave communications, such focus on the newly emerging cost-effective IRS. Generally as high signal attenuation and high vulnerability to blockage, speaking, M-MIMO based BSs can be deployed for a long- can be alleviated by exploiting the UAV mobility. For example, range coverage with massive devices, whereas mmWave and UAVs acting either as aerial platforms or users can not only fly IRS are more suitable for short-range coverage, especially towards ground nodes to reduce the propagation loss, but can when there are strong deterministic channel components, such also intelligently adjust their trajectories to bypass surrounding as the LoS path. Whereas both M-MIMO and mmWave require obstacles to increase the probability of LoS paths. However, the use of active antennas arrays and generally have higher similar to M-MIMO, mmWave communication requires the weight, hardware cost and energy consumption than IRS that use of a large number of antennas and the channel coherence uses passive reflecting elements. Besides, due to the passive time is shortened due to the high UAV mobility and shorter reflection mechanism of IRS, it requires a relatively larger wavelength signals. As such, fast channel variations will result aperture than M-MIMO and mmWave based active arrays in practice, which renders effective dynamic beam training and and also imposes new challenges on the related passive IRS tracking techniques imperative. Some existing works proposed channel estimation, which requires larger installation space in the use of movement prediction filters such as Kalman filters practice and may result in higher computational complexity. to track the time-varying UAV-ground channels [91]–[93]. Besides, a fast beam searching algorithm was developed in [94] to track the UAV-to-ground channels based on a predeter- A. M-MIMO and mmWave for UAV Communications mined hierarchical codebook. Other important topics for future M-MIMO, as a key enabling technology in the current research include low-complexity spectrum management [95], 5G standard, is promising for supporting cellular-connected and high-speed and reliable backhaul design. UAV communications [8], [77]–[81]. Equipped with full- dimensional large arrays, ground BSs can perform fine-grained 3D beamforming to mitigate interference among high-altitude B. IRS for UAV Communications UAVs and low-altitude terrestrial users, and thus, achieve Despite the promising advantages of M-MIMO and much higher network throughput. However, the beamforming mmWave communications, their required high complexity and gain of M-MIMO critically depends on the availability and hardware cost as well as increased energy consumption are still accuracy of channel state information (CSI) at the ground BSs, crucial issues faced in practical implementation [73], [74]. As whereas cellular-connected UAVs introduce new challenges. an alternative, IRS has recently emerged as a new and cost- First, UAVs can cause severe pilot contamination to a large effective solution to improve the received power and suppress number of ground BSs due to their strong LoS channels, air-ground interference in 3D space [96]–[99]. An IRS is which cannot be resolved by existing pilot decontamination composed of a large number of passive reflecting elements, techniques designed solely for terrestrial users. Second, the each of which is able to reflect the impinging electromag- UAVs’ high mobility in 3D space renders efficient beam netic wave with a tunable reflection coefficient (including tracking for them a more challenging task than for terrestrial an amplitude and a phase shift) [100]–[103]. By smartly users and can incur excessive pilot overhead [82], [83]. Third, coordinating the reflections of all elements, an IRS is able to practical implementations may adopt the hybrid beamforming reconfigure the wireless channel with the desired signals added based M-MIMO architecture to support a large group of coherently and interference cancelled at designated receivers, coordinated UAVs or a UAV swarm. This will further compli- thus significantly enhancing the communication throughput cate the pilot contamination and beam tracking problems. To without the need for deploying new active BSs or relays. better address the inter-cell air-ground interference to achieve Furthermore, IRSs possess appealing advantages in practice enhanced network throughput, a more ambitious approach such as light weight, and thus can be easily mounted on walls referred to as cell-free M-MIMO, which combines ideas and even the surface of high-speed moving vehicles to support from distributed/network MIMO and coordinated multipoint numerous applications [104]–[109]. As such, IRS has been transmission, was proposed recently, where massive antennas considered as a disruptive technology for transforming our cur- distributed over a large geographical area are connected with rent “dumb” radio environment into an intelligent one, which a central processing unit (CPU) [84]. In this case, both UAVs potentially benefits a wide range of vertical industries such and ground users are surrounded by multiple BS antennas, as transportation, manufacturing, and smart cities. Recently, rather than the conventional case where each BS is surrounded IRS has also been recognized as a promising technology for by multiple users [84]. Due to the LoS-dominant air-ground the future sixth-generation (6G) ecosystem [98], [110], [111] channels, such a user-centric architecture provides more de- and studied in various system setups [112]–[121]. Generally grees of freedom to exploit the macro diversity offered by the speaking, as shown in Fig. 2, IRS can be either deployed on many distributed BSs. Nevertheless, key issues that remain to the ground to assist UAV communications or attached to UAVs be solved include efficient centralized and distributed power to assist terrestrial communications [122], as elaborated below. control, low-complexity fronthaul/backhaul provisioning, and In addition, a brief comparison of existing works on IRS and network scalability with respect to UAV swarms [85]–[88]. UAV was summarized in Table III. 7

IRS

H

User 1 d User 2 D

(a) IRS aided UAV communications (b) UIRS aided terrestrial communications

Fig. 2. UAV communications with IRS.

1) Terrestrial IRS Assisted UAV Communications: For data multi-user systems. To address this issue, resource allocation, dissemination and collection in IoT networks, a single UAV including user scheduling, power allocation, UAV trajectory, may need to fly sequentially approaching each IoT de- and beamforming was optimized in [126] to minimize the vice/cluster for high data rate transmission. However, this not total power consumption of an IRS-assisted multi-user multi- only compromises the user access delay but also increases antenna UAV communication system, subject to a given re- the UAV propulsion energy consumption [21], [123], [124]. quired minimum achievable rate of each ground user. It was To overcome these drawbacks, one possible solution is to shown that deploying an IRS can shorten the UAV flying dispatch multiple UAVs, however this requires sophisticated trajectory without compromising the QoS, as compared to UAV-UAV coordination and thus increases the operational cost the case without IRS. This work was then further extended as well as the signalling overhead. As an alternative, a properly to multi-IRS scenarios [127], [128]. In particular, the over- deployed IRS can help resolve this issue. As shown in Fig. 2 all weighted data rate taking into account the geographical (a), the UAV only needs to directly cover a subset of the IoT fairness of all the users was maximized in [127] to jointly devices and the significant path loss between the UAV and optimize the UAV’s trajectory and the phase shifts of the other devices can be effectively compensated by leveraging reflecting elements of the IRSs. Instead of using conventional the deployment of IRS. This is particularly useful for uplink convex optimization methods, the authors in [127] proposed transmissions due to the limited energy supply of IoT devices, a low-complexity deep Q-network (DQN)-based solution by which helps reconcile the energy trade-off in ground-to-UAV discretizing the trajectory, which is useful for practical systems communications [25]. In addition, as mentioned previously, the with discrete phase-shift control. Then, they further proposed ground BS antennas are typically downtilted in current cellular a deep deterministic policy gradient (DDPG)-based solution networks, i.e., their main lobes point towards the ground for to facilitate continuous phase-shift control. optimizing the coverage for ground users, and UAVs flying In [129], the potential of IRS to enhance cellular communi- above the BSs are only supported through the side lobes. cations for UAVs was investigated, where the received signal Indeed, recent 3GPP studies have confirmed that UAVs receive power gain was analyzed as a function of the UAV height weak signals from existing terrestrial BSs and hence support- and various IRS parameters including its size, altitude, and ing aerial users will require further research and development distance from the BS. In particular, it was shown that for a [53], [54]. Fortunately, by leveraging the signals reflected by UAV hovering at 50 meters (m) above the ground, a 21 dB IRSs via 3D passive beamforming to serve the UAVs in the received power gain can be achieved with a properly deployed sky, the communication links between ground nodes and UAVs IRS comprising 100 passive elements based on the 3GPP can be greatly improved, which helps eliminate the need for ground-to-air channel models. Besides, due to the downtilted significantly modifying the configurations of existing ground antenna pattern of the BSs, the optimal IRS deployment alti- BS antennas. tude was shown to decrease as the IRS-BS distance increases so as to effectively reflect the signals coming from the BS. Motivated by the above advantages of IRS, there have been Also, in [130], an IRS was deployed to assist a UAV relaying a number of works in this research direction [125]–[130]. system where the average capacity, outage probability, and For instance, assuming that the UAV-ground link is blocked, average bit-error rate (BER) were analyzed. It was found that the authors in [125] maximized the average achievable data if the IRS is deployed between the source and the UAV relay, rate in IRS-assisted UAV-ground communication systems by the optimal position of the UAV for maximizing the outage jointly optimizing the UAV’s trajectory and the IRS phase performance moves closer to the destination, which is different shifts. It was shown that rather than flying towards the ground from the case without the IRS where the UAV relay needs to user, the UAV should fly towards the IRS to significantly be deployed in the middle between source and destination. improve the data rates by exploiting the reflections of the 2) UAV IRS Assisted Terrestrial Communications: In prac- large IRS aperture. Yet, this study only focused on the case tice, small lightweight UAVs usually can support only a limited of a single user being served by a single-antenna UAV and payload, and may not be able to carry bulky RF transceivers the proposed solution was not applicable to more general in BSs/relays. Besides, such active nodes generally incur 8

TABLE III COMPARISONOFEXISTINGWORKSON IRS AND UAV.

IRS use case Reference No. of IRS Design Objective Approach [125] Single Rate maximization AO and SCA [126] Single Power minimization AO, SCA, and Lagrange duality Terrestrial IRS [127] Multiple Weighted rate maximization Reinforcement Learning [128] Multiple Received power maximization AO and SCA [131] Multiple BER minimization Penalty based algorithm [122] Single Minimum SNR maximization Two-step method [132] Single Rate maximization Reinforcement learning UIRS [133] Single Secrecy energy efficiency AO and SCA [134] Single EE maximization Fractional programming

high energy consumption, which further aggravates the energy where P denotes the BS transmit power, β0 denotes the consumption of UAVs and limits endurance. Furthermore, in channel power gain at a reference distance of 1 m, N denotes terrestrial wireless networks, the deployment of IRS is highly the number of IRS reflecting elements, and σ2 denotes the restricted by the availability of the existing infrastructure such receiver noise power. By setting the first-order derivative of as building facades, lamp posts, and advertising boards. As the denominator with respect to d to zero, the optimal UIRS such, unlike the case in Fig. 2 (a), it may not be possible to placement for maximizing the receive SNR can be easily find proper deployment locations for installing an IRS to es- derived as [102], [122] tablish the desired (virtual) LoS links between the transceivers. √ D ± D2 − 4H2 Compared to terrestrial IRS, a UAV-based IRS (UIRS) is more  , for D ≥ 2H, d∗ = 2 (2) likely to have strong LoS links with the ground nodes due D to the relatively higher altitude and mobility of UAVs, thus  , otherwise. 2 reducing the probability of a blockage between them [135]. This is particularly important for high frequency mmWave It is interesting to note that depending on the values of H and and THz communication systems that are very sensitive to D, the optimal deployment location changes from the midpoint blockage [132]. For example, as shown in Fig. 2 (b), since an to locations with equal distances to the transmitter (BS) and IRS only reflects the signals to its front halfspace, deploying an the receiver (user 2), as also verified in Fig. 3 for P = 20 dBm, 2 IRS on the building facade only helps to enhance the commu- β0 = −30 dB, N = 250, σ = −100 dBm, and D = 400 m. nication data rate of user 1 and the lack of coverage behind Considering that D in practice is expected to be considerably the building remains still unsolved. Although it is possible larger than H to obtain effective reflection from the UIRS, to connect user 2 to the BS via multi-hop IRS reflections the UIRS should be deployed close to the transceivers, which if multiple IRSs are deployed, this requires properly located is quite different from the deployment strategy for active buildings in the surrounding environment as well as additional relays. Nevertheless, practical UIRS deployment also needs multi-IRS coordination and signal processing complexity. In to consider other factors such as LoS versus non-LoS links, this case, deploying a UAV equipped with an IRS is practically channel rank and condition number, and the use of a single appealing [122]. More importantly, UAVs with high mobility large UIRS or multiple smaller cooperative UIRSs [98], [100], are able to adapt their positions dynamically according to [136]. Furthermore, the physical UIRS elements orientation the changes in the communication environment and thus can also has a great effect on the air-ground channel characteristics maintain persistent (virtual) LoS links between transceivers. and hence the system performance, which calls for further To provide some insights on UIRS deployment, in Fig. 2 (b), investigation in future. we focus on the UIRS-aided communication for user 2 and the However, there has been only limited work in this line of BS-user distance is D m. The BS-UIRS horizontal distance research so far [132]–[135], [137]. In [135], the minimum SNR is d m and the UIRS’s altitude is H m. We optimize the within a given rectangular service area (e.g., a hot spot in a cel- horizontal location of the UIRS (d) to maximize the achievable lular network) was maximized by jointly optimizing aerial IRS rate/signal-to-noise ratio (SNR) of user 2. For simplicity, LoS placement, the IRS phase shifts, and the transmit beamforming channels are assumed for both the BS-UIRS and UIRS-user at the BS. To gain useful insights, the authors first derived links, while the BS-user direct link is assumed to be severely the optimal UIRS placement and phase shifts for the special blocked and thus ignored. Then, the receive SNR at user 2 case of single-location SNR maximization, where the optimal assuming optimal IRS passive beamforming is given by [102], horizontal location was shown to depend only on the ratio [122] between the UIRS height and the source-destination distance. As for the general case of area coverage enhancement, an P β2N 2 SNR = 0 , (1) efficient two-step method was proposed by decoupling the (d2 + H2)((D − d)2 + H2)σ2 phase-shift optimization from the UIRS placement design, 9

the development of robust algorithms for tuning the phase 10 shifts at the IRS is still a practically challenging task. Joint H = 20 m 9 H = 50 m reflection amplitude and phase-shift control at the IRS could be H = 100 m a possible approach to alleviate air-ground interference, which, H = 250 m 8 however, increases the hardware cost and design complexity. Hence, how to strike an optimal performance-complexity/cost 7 tradeoff when considering practical hardware imperfections

6 such as discrete and/or coupled reflection amplitudes and phase shifts [139]–[142] remains an open problem. In addition, 5 practical low-complexity channel estimation/tracking methods

Achievable rate (bps/Hz) are required for acquiring the UIRS-user channels along their 4 3D trajectories, which is more challenging than the case of 3 terrestrial IRSs that are at fixed locations [98]. 2) Deployment of IRS/UIRS and UAV-IRS Symbiotic Sys- 2 0 50 100 150 200 250 300 350 400 tems: For a given network with both ground users and BS-UIRS horizontal distance, d(m) UAVs, how to jointly optimize the resource allocation and the deployment of terrestrial IRSs and UIRSs including their locations and densities is another important and interesting Fig. 3. Achievable rate versus UIRS placement. topic to pursue. Finally, most existing studies focus on the utilization of IRSs to enhance the primary end-to-end com- where a 3D beam broadening and flattening technique was munication by exploiting passive beamforming [125]–[130], proposed to obtain 3D beam patterns matching the service [132]–[135], [137]. However, in practice, UIRSs also need area size on the ground [122]. For mobile users in practical to deliver their own information, including control signals systems, the dynamic self-positioning of UIRS is required, to acknowledge their current status, environmental parameters which motivated the work in [132]. Specifically, to maximize (such as temperature, humidity and pressure) obtained by UAV the downlink mmWave communication capacity of a mobile sensors, etc., to the BS. Rather than equipping each UIRS with outdoor user, a reinforcement learning (RL) approach based a dedicated transmitter, which is not cost-effective and also on Q-learning and neural networks was proposed in [132] to requires extra power consumption, a more appealing approach enable efficient deployment and phase-shift optimization for is to modulate the IRS information onto the reflected signals a self-sustained UIRS. The main advantage of this approach to achieve passive information transfer, which leads to the new lies in that it requires no prior knowledge of the dynamic paradigm of symbiotic communication [131], [143], [144] that environment, but learns the characteristics of the environment aims to simultaneously transmit IRS information and enhance during the service process of the UIRS, based on data mea- the communication quality of the primary link. surements and feedback received during each communication stage. The results showed that significant gains can be achieved III.ULTRA-RELIABLEAND LOW-LATENCY by using such a UIRS as compared to a static IRS in terms COMMUNICATIONS of the average data rate as well as the LoS probability. Yet, to simplify the system design, it was assumed in [132] that the To fully embrace the era of the Internet-of-Everything BS only transmits, and thus, the UIRS only reflects signals (IoE), a well-developed wireless infrastructure focusing on low during the hovering state. latency and high reliability is necessary. As a result, starting from the 5G wireless systems, the concept of URLLC has been introduced as a key performance indicator [147], [148]. Unlike C. Future Research classical high data rate oriented multimedia streaming services, Despite these existing works, the study of IRS-assisted in which high rate data flows from a source to a sink, typical eMBB air-ground communications is still in a very early URLLC services focus on conveying sensing information, stage and many interesting and important problems need to control commands, and feedback information in short packets be addressed. which need to be delivered reliably in an extremely short 1) Joint UAV Trajectory and IRS Reflection Pattern Design: period. Table IV shows some emerging URLLC applications First, the IRS reflections could be very sensitive to UAV and requirements. It can be seen that these applications impose jittering and user location uncertainty [138], which may cause heterogeneous requirements on both the reliability and latency. significant performance degradation for UAVs and/or ground In particular, for the Internet-of-Drones (IoD), the need for users due to the reliance on LoS channels. Particularly, the cur- rapidly conveying control signals in highly dynamic UAV rent joint UAV trajectory and IRS phase-shift designs utilized communication scenarios imposes strict requirements on both trajectory discretization [125]–[128] where the UAV is con- latency and reliability. Unfortunately, despite the development sidered to be at a fixed sample location during each interval. of advanced wireless communication techniques, e.g., massive However, in reality, the UAV is continually moving, which can MIMO [149]–[152], mmWave transmission [153]–[155], and easily render the IRS passive beamforming ineffective due to full-duplex communication [156]–[158], it is believed that a the signal misalignment arising from UAV movement. As such, single wireless link between a user and a terrestrial BS cannot 10

TABLE IV APPLICATIONS AND REQUIREMENTSFOR URLLC IN 5G-AND-BEYOND SYSTEMS [145]–[147].

Applications Latency (ms) Reliability (%) Data Size (bytes) Communication Range (m) Smart Grid 3 ∼ 20 99.999 80 ∼ 1000 10 ∼ 1000 Professional Audio 2 99.9999 3 ∼ 1000 100 Automated Vehicles 1 99 144 400 E-Health 30 99.999 28 ∼ 1400 300 ∼ 500 Argument Reality 0.4 ∼ 2 99.999 12, 000 ∼ 16, 000 100 ∼ 400 Intelligent Transportation Systems 10 (end-to-end) 99.9999 50 ∼ 200 300 ∼ 1000 Vehicle-to-vehicle (V2V) 5 99.999 1600 300 Tactile Internet 1 99.99999 250 100k Internet-of-Drones (Video) 1 ∼ 10 99.999 1 − 20, 000 Ground-to-air ∼ 400 satisfy URLLC requirements. As a result, different kinds of Next, we show how to leverage short-packet communication diversities are necessary to offer the flexibility to trade more to meet the requirement of URLLC applications. bandwidth for achieving shorter delays and higher reliability [159]. A. Short-Packet Communication Recently, UAV-based communications have been widely To guarantee URLLC in UAV-based communication sys- recognized as one of the disruptive and enabling technolo- tems, various approaches have been proposed in the literature gies to address this issue. Indeed, aerial nodes usually en- and 5G standardization. For instance, by exploiting the exces- joy unobstructed LoS air-to-ground channels [160]. Thus, sive spatial degrees of freedom offered by massive MIMO, conventional terrestrial communication systems with blocked in theory, the reliability of UAV-based communications in the communication links may deploy UAVs as aerial relays to ground-to-air links can be theoretically improved to a certain substantially improve system data rate and/or coverage to extent [162]. However, the deployment of a large number of realize URLLC by exploiting spatial diversity. Besides, in the antennas at UAVs is challenging due to the aforementioned case of natural disasters and disease outbreaks, it may not energy and size limitations of practical UAVs. The actual com- be possible to guarantee URLLC via standalone conventional munication latency, TL, is comprised of several components terrestrial communication networks. In these cases, UAVs which can be summarized as follows: can be employed as ad-hoc aerial BSs to offer temporary T = T + T + T + T + T , (3) URLLC links. In fact, by exploiting the high maneuverability L ttt prop proc retex sig of UAVs, fast, highly flexible, and cost-effective deployment of where Tttt denotes the time-to-transmit latency, which is the communication infrastructure can be ubiquitously established, time needed to transmit a packet; Tprop is the signal propaga- especially in temporary hot spots, disaster areas, complex ter- tion delay from the transmitter to the receiver; Tproc is the time rains, and rural areas. On the other hand, UAVs can also serve consumed for encoding, decoding, synchronization, and chan- as a physical carrier for user equipment connecting to existing nel estimation; Tretex is the time required for retransmission networks requiring URLLC services, e.g., to report sensing (if any); and Tsig is the time incurred by signaling exchanges information. In both cases, physical communication of UAVs such as connection request, scheduling grant, channel training, can be divided into payload delivery (PD) and control and non- feedback, and queuing delay. In response to the harsh maxi- payload communication (CNPC) [8]. Specifically, for CNPC mum delay allowance specified in Table IV, different methods links, short-packet control information is required to be con- for reducing the overall communication latency have been veyed from a ground BS to UAVs or exchanged between UAVs proposed in the literature [163] by minimizing the different with high reliability and low latency [161]. As for PD links, components in (3). For instance, if the communication link UAVs can be adopted to provide certain types of data services is shadowed, one can deploy a UAV as a wireless relay to requiring not only extremely high data rates but also ultra-low establish an additional reliable end-to-end communication link latency and high reliability, e.g. conveying real-time video and which can help to reduce the time spent on retransmission, haptic information in the Tactile Internet. However, existing i.e., Tretex. Also, non-coherent transmission and detection wireless communication protocols and techniques were de- were advocated in [164]. In particular, for these schemes, signed mainly for terrestrial communications and do not fully acquiring the knowledge of instantaneous channel conditions capture the hybrid characteristics of air-to-ground and air-to- is not necessary, which reduces both Tproc and Tsig needed air communications. More importantly, applying conventional for channel training and estimation at the expense of some techniques/protocols to UAV-based communication systems performance loss in signal detection. Besides, a new short may introduce an exceedingly long delay and raises serious slot structure exploiting the concept of mini-time slots was safety issues for controlling aerial nodes, which remains a proposed for 5G [165] to realize URLLC in practice via major obstacle for realizing efficient UAV communications. shortening Tttt in (3). 11

On the other hand, to further improve system performance, channel capacity-achieving error-correcting codes have been proposed and adopted in 5G communication systems, e.g., UAV polar codes and low-density parity-check (LDPC) codes. How- h h 1 relay 2 ever, due to the requirement of short packets with low latency, the blocklength of the coded messages is short which in turn jeopardizes the communication reliability. More importantly, the Shannon capacity, which assumes infinite blocklength codes, is no longer achievable and cannot be used to accu- rately characterize the system performance. Although one can User perform proper resource allocation to reduce the performance Blockage degradation by taking into account the impact of short block- BS length codes, most existing resource allocation algorithms for UAV communications, e.g., [166]–[169], were designed based Fig. 4. An example of adopting a UAV as a wireless relay taking into account on Shannon capacity as the performance metric, thereby as- the impact of finite blocklength. suming implicitly infinite block length transmission. Besides, these algorithms do not offer sufficient flexibility to strike a balance between delay (codelength) and reliability. As can be observed from (4), the subtrahend accounts for the penalty caused by the finite blocklength. In particular, Recently, driven by the need for URLLC, the study of for a sufficiently long blocklength, the normal approximation SPC has received significant attention in the literature [170]– approaches the conventional Shannon capacity. As a result, [173], as summarized in Table V. These works have captured the normal approximation in (4) serves as a useful tool for the relationship between the achievable rate, decoding error optimizing the allocation of the system resources to strike probability, and packet length, which facilitates the design of a balance between the blocklength (delay), achievable rate, effective resource allocation. In the following, we summarize and reliability. In the following, we focus on an illustrative the key techniques for resource allocation design taking into example to highlight how URLLC can be facilitated in UAV account the impact of short packets in UAV communications. communication systems via resource allocation. 1) Performance Metric for SPC: Unlike conventional sys- tems, where Shannon’s capacity theorem can be used to 2) Example for Resource Allocation for URLLC: Suppose characterize performance and the probability of erroneous there is a ground BS communicating with a ground user. Due decisions becomes negligible as the coded packet length is to heavy shadowing in the direct channel, as shown in Fig. 4, a sufficiently long, short packets are adopted in URLLC systems UAV adopting the decode-and-forward protocol is dispatched to achieve low latency, which makes decoding errors unavoid- to serve as a wireless relay to facilitate reliable end-to-end able. As an alternative, for evaluating the performance of SPC, communication resulting in a two-hop relaying system [176], the notion of normal approximation for finite blocklength [177]. The 3D coordinates of the BS, the user, and the UAV codes was developed in [174]. For ease of illustration, we are (0, 0, 0), (a, b, 0), and (x, y, H), respectively, with H being focus on a simple UAV-to-ground BS communication channel. the UAV’s fixed altitude. For ease of presentation, we denote Suppose that the BS transmits a message to the UAV over a the channel power gain from the BS to the UAV and that quasi-static flat fading channel with a given packet length, from the UAV to the user by h1 and h2, respectively. Also, the BS N = WT , where W is the bandwidth and T denotes the transmit powers of the BS and the UAV are given by P and B UAV transmission duration. The achievable rate1 (bit/s/Hz) for the P , respectively. It is well known that for a sufficiently high given finite blocklength can be accurately approximated as altitude, the free-space channel model can accurately capture follows [174]: the channel characteristic. Thus, h1 and h2 are modelled as

r −1 V Q () β0 β0 R ≈ C − , (4) h1 = 2 2 2 , h2 = 2 2 2 , (6) NB ln 2 H + x + y H + (a − x) + (b − y) where C = log2(1+γ) denotes the Shannon channel capacity respectively, where β0 is the channel power gain at a given that can be achieved for infinite packet lengths, γ is the reference distance. As a result, the received SNR at the UAV received SNR which is a function of the UAV’s trajectory, and and the user can be expressed as  is the required decoding packet error probability. Variable V is the channel dispersion and Q−1(·) is the inverse of the Gaus- PBSh1 PUAVh2 1 R ∞  t2  sian Q-function, where Q(x) = √ exp − dt. For γ1 = 2 , γ2 = 2 , (7) 2π x 2 σUAV σUser the complex-valued additive white Gaussian noise (AWGN) channel, the channel dispersion is given by 2 2 respectively, where σUAV and σUser denote the power of the AWGN at the UAV and the user, respectively. V = 1 − (1 + γ)−2. (5) As can be observed from (7), the received SNRs of the two 1Based on the results in [175], the approximation error in (4) is 0.1 − 2 hops depend on the position of the UAV. Then, by applying dB for decoding error probabilities from 10−3 to 10−7. the normal approximation in (4), the achievable rates of the 12

first hop and the second hop are given by 1.8 r −1 blocklength, P = 33 dBm V1 Q () BS Finite blocklength, P = 33 dBm R1(m1, x, y) = log2(1 + γ1) − , (8) 1.6 BS m1 ln 2 blocklength, P = 23 dBm BS r −1 Finite blocklength, P = 23 dBm V2 Q () 1.4 BS R2(m2, x, y) = log (1 + γ2) − , (9) blocklength, P = 13 dBm 2 m ln 2 BS 2 Finite blocklength, P = 13 dBm 1.2 BS respectively, where m1 and m2 are the blocklengths adopted at the BS and the UAV, respectively. Furthermore, the channel 1 dispersions of the first hop and the second hop are given     0.8 −2 −2 by V1 = 1 − (1 + γ1) and V2 = 1 − (1 + γ2) , 0.6 respectively. Achievable data rate (bits/s/Hz) 0.4 In practice, it is common to maximize the total achievable rate of the user by jointly optimizing the position of the UAV 0.2 relay and the blocklengths in the two hops, taking into account 0 the end-to-end communication delay requirement. Thus, the 0 100 200 300 400 500 proposed resource allocation policy is determined by solving UAV position (meters) the following optimization problem: ( ) 1 Fig. 5. A comparison of the achievable data rate of a UAV-assisted relaying maximize min R1(m1, x, y),R2(m2, x, y) network for different BS transmit powers and different blocklengths. The m1,m2,x,y 2 transmit power of the UAV hovering at a fixed altitude H = 100 m is 23 dBm. A ground user is located 500 m away from the BS. s.t. C1: (m1 + m2)TBlock ≤ Dmax, + C2: mi ∈ Z , i ∈ 1, 2, 2 2 2 blocklength case is obtained by optimizing m and m in (10). C3: x + y ≤ rmax, (10) 1 2 For comparison, we also consider a performance benchmark where TBlock in (10) denotes the time needed for conveying for which the delay constraint C1 in (10) is dropped and an one unit of blocklength and constraint C1 guarantees that infinite blocklength is assumed for transmission. As can be the maximum transmission delay does not exceed Dmax. observed from Fig. 5, there is a performance gap between Constraint C2 enforces that the blocklengths are non-negative the cases of finite blocklength and infinite blocklength, as the integers. Finally, constraint C3 ensures that the position of the former has an insufficient number of coded information blocks UAV relay is within the serving cell with a radius of rmax. for sufficiently averaging out the impact of Gaussian noise to Problem (10) is a non-convex optimization problem. The approach the system capacity. This gap is generally inevitable non-convexity is caused by the combinatorial constraint in in URLLC as the end-to-end delay constraint is stringent. C2 and the non-convex normal approximation appearing in Furthermore, the optimal UAV position maximizing the end- the objective function. However, by analyzing the structure of to-end achievable rate depends heavily on the power budgets problem (10), it can be solved optimally via a combination of of both the UAV and the ground BS. In fact, the optimal UAV the penalty method and monotonic optimization [178] at the position attempts to balance the achievable rate of the two hops expense of a high computational complexity. As an alternative, in the system. In particular, when the power budget at the BS is one can apply approximations to simplify the problem at hand. smaller than that of the UAV, e.g. PBS = 13 dBm, PUAV = 23 For instance, a high SNR approximation is commonly adopted. dBm, the optimal UAV position is closer to the BS than to the In particular, in the high SNR regime, i.e., γ → ∞, the channel user. Although the power budget of the BS is limited, the dispersion can be approximated by optimal strategy reduces the path loss between the BS and 1 the UAV for establishing an efficient first communication hop. V = 1 − ≈ 1 (11) (1 + γ)2 Besides, the higher UAV transmit power can be exploited to compensate for the longer communication distance between such that it becomes a constant. Then, successive convex the UAV and the ground user to balance the achievable rate approximation and the penalty method can be employed to of the two hops. In contrast, when the power budget of the obtain a suboptimal solution to (10) with polynomial time BS is larger than that of the UAV, e.g., P = 33 dBm and computational complexity. BS P = 23 dBm, the optimal UAV position moves towards In Fig. 5, we show the achievable data rate of the considered UAV the user, allowing the system to again fully exploit the power UAV-assisted relaying system versus the position of the UAV budgets of both the BS and the UAV to maximize the system for different BS transmit powers. The maximum delay is 1 performance. ms while each mini-time slot is TBlock = 0.01 ms with 200 kHz of bandwidth. In other words, there are a maximum of 100 blocks for transmission. For simplicity, the channel-to- B. Future Research noise ratio (CNR) at the reference distance of 1 meter is set 1) Multiple Access: A critical issue in realizing URLLC β0 β0 as 2 = 2 = 60 dB and the required decoding packet in UAV systems is the limited spectrum available for simul- σUser σUAV error probability is  = 0.1%. The achievable rate for the finite taneous ground-to-ground, ground-to-air, and air-to-air com- 13

TABLE V COMPARISONOFEXISTINGWORKSON URLLC IN UAV COMMUNICATIONS.

Reference Methodology Objective Advantage Disadvantage [164] Non-coherence transmis- Reducing Tproc and Tsig Easy to implement Sensitive to decoding error sion and detection [165] Introducing mini-time slot Shortening Tttt Achieving low latency Protocol update is needed [179] Short error-correcting Reducing Tproc and Tsig Striking a balance be- Difficult to design efficient code codes tween latency and error structure and decoder rate [170] Joint UAV altitude and Maximizing available UAV communi- Enjoying LoS and adap- Challenging optimization problem transmission duration op- cation range subject to delay constraints tive to actual environment timization [171] Joint blocklength Minimizing end-to-end communication Enjoying LoS and adap- Challenging optimization problem and UAV variable error probability subject to blocklength tive to actual environment optimization constraints [173] Resource allocation for Maximizing the achievable rate for a Adaptive to actual envi- Challenging optimization problem short packet communica- given finite blocklength ronment tion

TABLE VI COMPARISONOFEXISTINGWIRELESSSYSTEMSFORSUPPORTING IOE [180], [181].

Zigbee Bluetooth WiFi LoRa Cellular Spectrum Unlicensed Unlicensed Unlicensed Unlicensed Licensed Connectivity Moderate Small Large Massive Massive Throughput Moderate Low High High High Coverage Short Short Moderate Long Long Security level Moderate Low Moderate High High Power consumption Low Low High Low Low Mobility No No No Yes Yes Cost Low Low Low Moderate High munications. Unlike terrestrial communication systems where 2) Hardware Imperfections: Besides, existing algorithms ground BSs are connected to a data hub via high-speed fixed- for URLLC in UAV-based communication systems have been line backhaul links, e.g., optical fibres, the implementation of developed based on the assumption of perfect hardware. Yet, ground-to-air and air-to-ground links usually relies on dedi- various types of hardware imperfections exist in practical UAV cated wireless communication channels. In particular, UAV- systems and must be accounted for [182]. For example, due to based communication systems have a stringent demand for the finite precision of electronic circuits and the imperfection system resources in both the ground-to-air and air-to-ground in manufacturing mechanical components, the control of the links due to the required support of high data rate backhauling UAV’s trajectory and positioning may not perform as expected and exchange of time-critical UAV control signals. In this [138]. As for the hardware modules for wireless commu- context, the allocation of dedicated orthogonal radio resources nication, phenomena such as power amplifier non-linearity, to ground-to-air links is commonly considered in the literature. non-linear phase noise, frequency and phase offsets, in-phase Yet, the spectral efficiency of orthogonal resource allocation and quadrature (I/Q) imbalance, and quantization noise jointly is low. In fact, the wireless spectrum is scarce and is already degrade the decoding capability of the receivers. Hence, it is congested by existing communication systems. Furthermore, in still unclear whether it will be possible to guarantee URLLC order to guarantee URLLC, multiple UAVs should be deployed in UAV-based communication systems. Thus, a thorough per- simultaneously which puts a significant burden on the need formance analysis taking into account the impact of hardware for available spectrum. Hence, conventional multiple access impairments is needed. schemes based on orthogonal spectrum partition would quickly exhaust the available resources even for moderate numbers IV. MASSIVE MACHINE-TYPE COMMUNICATIONS of UAVs and ground users. More importantly, this would The connection density of MTC networks will be around 10 introduce a long delay in the provisioning of URLLC and million devices per km2 by 2030 [183]. Currently, for fast and raises serious safety issues in controlling aerial nodes. Thus, cheap implementation, machine-type devices (MTD) access efficient multiple access schemes based on non-orthogonal wireless networks via low-cost commercial technologies, e.g., spectrum utilizations have to be developed for enabling UAV- Zigbee, Bluetooth, and WiFi. Besides, the application of long based URLLC systems. range radio (LoRa) and cellular networks to the IoE has been proposed for enabling massive access. Indeed, it is beneficial 14 and economical to deploy the IoE using existing networks. by taking into account the periodic and sparse transmission However, traditional communication networks are designed characteristics of sensors, UAV semi-persistent scheduling can and used for human-to-human (H2H) communications to serve be performed to reduce control plane traffic between the UAV a relatively small number of users, compared to the large and the BS. On the other hand, as an MTD itself, a UAV number of devices in IoE, cf. Table VI. The fundamental can help relay the data from the ground MTDs so as to challenges in the implementation of MTC are massive connec- improve the energy efficiency of the whole communication tivity, ultra-high reliability and low latency requirements, and system. However, despite the various practical advantages of energy-efficient transmission. As URLLC has been discussed deploying UAVs, the performance of UAV-enabled MTC is in Section III, in this section, we tackle the challenges in still limited by the small onboard battery equipped at UAVs MMTC from the perspectives of massive access and energy- due to the size, weight, and power (SWaP) constraints of efficient transmission. UAVs. Hence, practical methods for enabling sustainable MTC Unlike traditional communication networks which focus on via UAV communications are needed. the performance of the downlink for a small number of users, MTC has a much higher demand on system resources in the A. NOMA uplink due to the associated massive uplink connectivity [184]. For instance, security cameras can be installed on UAVs for NOMA is a key enabling technology for future mMTC capturing images of specific areas for security surveillance. applications due to its capability of allowing simultaneous The captured images are periodically conveyed to a ground BS transmissions of multiple devices in the same resource block for further data analysis. In this context, allocating dedicated [148], [180], [192]. Specifically, NOMA leverages superposi- radio resources to MTD orthogonally has been proposed in tion coding (SC) at the transmitters and successive interference the literature, e.g., [23], due to the simple design. However, cancellation (SIC) at the receivers to achieve efficient access for MTC, since the number of MTDs is large, orthogonal as well as to partially mitigate co-channel interference. It has multiple access (OMA) would quickly exhaust the valuable been shown that NOMA is particularly effective for cases radio resources and introduce exceedingly long delays. Fur- where the users experience substantially different channel thermore, some commonly adopted random access protocols conditions [192]. for H2H communications such as Carrier Sense Multiple 1) NOMA for Cellular-connected UAVs: Due to its ad- Access (CSMA) and ALOHA are strictly suboptimal for MTC, vantages, NOMA is practically appealing for application in since they suffer from congestion and overloading in the cellular networks with co-existing UAV and ground users. presence of a huge number of MTDs. Therefore, efficient Particularly, NOMA allows UAVs/a UAV swarm to reuse the multiple access mechanisms such as grant-free random access resource blocks that are already occupied by ground users, thus and NOMA [185]–[187], are necessary to enable the massive improving the number of aerial users that can be supported connections required for MTC communications. In addition, even for a high ground user density, as compared to the MTDs for IoE are usually powered by batteries with limited non-scalable OMA scheme. On the one hand, more reliable capacities in practice. Although the lifetime of the traditional air-to-ground communications can be established due to the H2H communication networks can be extended by replacing or existence of LoS links, as compared to the non-LoS terrestrial recharging the batteries, this may be inconvenient, dangerous channels between the ground users and BSs. Besides, the (e.g. in a toxic environment), and costly in MTC networks due strong LoS air-ground channels allow the UAVs to be visible to to the large number of MTDs. Generally speaking, the energy a large number of ground BSs over a wide area. As a result, the consumption challenge to achieve green and self-sustainable UAVs can be potentially served by more BSs simultaneously MTC networks could be addressed by a proper combination than the ground users, thus achieving a higher macro-diversity of energy harvesting and/or energy-efficient designs which has gain in terms of user association. Besides, using NOMA drawn significant interest from both academia and industry in such unique air-ground channel environments also brings [73], [74], [150], [188]–[190]. challenges since the UAVs’ uplink transmissions will greatly The emergence of UAVs as a viable solution to deliver degrade the received ground users’ signals at a large number communication services in the past decades has introduced of ground BSs, considering the high frequency reuse factors a paradigm shift in MTC. Conventionally, long-distance com- of current multi-cell systems. Furthermore, ensuring that each munication between ground MTDs and their home BSs creates individual BS can cancel the UAV’s uplink interference via a system performance bottleneck which results in a strict limit signal decoding may severely limit the performance gain of on the lifetime of the communication network. In fact, by NOMA over OMA, since the UAV’s achievable rate will be exploiting the flexility and high-mobility of UAVs, one can then limited by the BS with the worst channel conditions. deploy a UAV as a sink node to collect data directly from To address this issue, a new decode-and-forward (DF) based ground MTDs. For example, for the case of a large sensor cooperative NOMA scheme, was proposed in [193], which network deployed in a rural area with limited cellular cover- exploits interference cancellation among collaborative adjacent age, a UAV can serve as a sink node to collect information BSs with backhaul links, e.g., the existing X2 interface in LTE data from the sensors [191]. In particular, by exploiting the [194]. Specifically, some BSs with better channel conditions LoS communication channels from the UAV to the ground are selected to decode the UAV’s signals first, and then sensors, the sensors can transmit with low power which can forward the decoded signals to their backhaul-connected BSs help extend the lifetime of the sensing network. Furthermore, for interference cancellation. It was shown that the proposed 15 scheme achieves higher data rates than both OMA and non- and/or long flight duration, NOMA and time division multiple cooperative schemes, especially when the ground traffic is access (TDMA) achieve the same rate performance. How- congested. To further improve the performance, a quantize- ever, NOMA generally outperforms OMA (including TDMA and-forward (QF) based cooperative interference cancellation and frequency division multiple access (FDMA)), and the approach was proposed in [195] where the adjacent BSs only capacity gain achieved by NOMA over OMA decreases as quantize the received UAV signals without decoding them. the UAV maximum speed and/or flight duration increases. In contrast, in the downlink, UAV receivers suffer from The comparison of two-user NOMA and OMA for UAV- strong co-channel interference from a large number of ground assisted communication was also extended to other design BSs. However, the interference mitigation techniques proposed objectives such as sum-rate [199] and outage probability [200]. for uplink transmission are not applicable in the downlink In particular, [200] considered a more practical Rician air- because the roles of the UAVs have changed from interference ground channel setup where the UAV flies with a constant sources to interference victims. One straightforward approach speed following a circular trajectory, with the objective of is to increase the transmit power of all ground BS. However, minimizing the outage probability. The condition under which this provides only a marginal gain as the interference increases NOMA outperforms TDMA was also derived in terms of the as well. Another possible practical approach is to leverage channel and UAV trajectory parameters. cooperative beamforming, where the available BSs that are not For a general setup with multiple users, proper user pairing serving ground users in the UAV’s occupied resource block, with bandwidth allocation is also important to unlock the full transmit collaboratively to the UAV to improve its received potential of NOMA. In [201], the UAV was assumed to pair power to overcome the co-channel interference. However, as one near user (cell-centered) with one far user (cell-edge). the ground user density increases, the number of such BSs Then, the multi-user rate max-min optimization problem was drops quickly and the co-channel interference also increases, formulated by jointly optimizing the bandwidth allocation, thus rendering this approach ineffective. To overcome this power allocation, UAV altitude, and antenna beamwidth. difficulty, a new cooperative beamforming scheme with in- In subsequent works, the use of NOMA was extended to terference transmission and cancellation (ITC) was proposed multiple-antenna and multi-UAV networks [202]–[206]. In in [196]. Specifically, the signals for terrestrial users reusing [202], the downlink transmission from a multi-antenna UAV to the same resource block as the UAV are first forwarded multiple ground user clusters was considered where analytical by their home BSs to the BSs serving the UAV, and then expressions for the outage probability and the ergodic rate transmitted along with the UAV’s signals via cooperative were derived. This was then extended in [203] by considering beamforming. As a result, the desired signal power at the the use of multiple UAVs in a large-scale cellular network. UAV receiver is improved and the terrestrial interference is In [205], the user angle was exploited as feedback infor- also suppressed, without affecting the existing transmissions. mation for mmWave NOMA communications. Compared to This centralized implementation requires excessive backhaul the conventional limited feedback scheme based on users’ transmissions among different BSs, which may be costly distances, angle information was shown to have a significant to implement. To lower the implementation complexity and potential in providing better separation for NOMA users in the signaling overhead, a distributed algorithm requiring only local power domain specifically for scenarios with multi-antenna information exchange among BSs was proposed based on the transmission. In [206], the resource allocation problem in a concept of divide-and-conquer in [196]. It was demonstrated multi-UAV aided IoT NOMA uplink transmission system was that such a distributed design can still significantly improve the studied where the channel assignment, the uplink transmit UAV’s performance as compared to the conventional schemes power, and the flying heights of UAVs were jointly optimized without ITC, especially when the terrestrial user density is to maximize the system capacity. high. However, current research results only consider a limited number of UAVs and the support of massive UAVs or UAV swarms is still a challenging problem that needs further efforts B. Energy Harvesting and Energy-efficient Designs [197]. Besides multiple access schemes, promising techniques 2) UAV-enabled NOMA: By proactively leveraging their to prolong the lifetime of MTDs have also been proposed high mobility, UAVs as aerial BSs inherently possess the including energy conservation schemes and wireless power ability to effectively exploit asymmetric channel conditions transfer (WPT) techniques [207]–[213]. Particularly, WPT of different ground devices to realize the performance gains provides several competitive advantages over the other energy promised by NOMA. A comparison of existing works is shown harvesting techniques such as its continuous availability and in Table V. For instance, the authors in [198] studied UAV long service range. However, the performance of energy har- enabled NOMA transmissions to two quasi-static ground users. vesting at the MTDs critically depends on the locations of The capacity region was characterized by jointly optimizing the power sources (e.g., BS, WiFi, etc.). For MTDs that are the UAV’s trajectory and transmit power/rate allocations over close the power sources, a sufficient amount of energy can be time, subject to the constrained maximum speed and transmit harvested, conversely, for those which are far away from the power of practical UAVs. A fundamental result is that regard- power sources, energy harvesting may not be possible at all less of the multiple access scheme, the optimal UAV trajectory due to the high sensitivity of typical energy receivers [73], follows a simple hover-fly-hover (HFH) policy. In particular, [214]. Fortunately, UAVs with their high mobility provide it was revealed that in the cases of infinitely high UAV speed a new promising solution to tackle these shortcoming of 16

TABLE VII COMPARISONOFEXISTINGWORKSON NOMA AND UAV.

Reference System Setup Air-Ground Channel Model Design Objective Approach [198] Single UAV, two users LoS channel Capacity region Closed-form expression [199] Single UAV, two users Probabilistic LoS channel Sum rate maximization Gradient descent Outage probability [200] Single UAV, two users Rician channel Closed-form expression minimization [201] Single UAV, multiple users LoS channel Minimum rate maximization SCA [204] Single UAV, multiple users LoS channel Sum rate maximization Two-step method [205] Multiple UAVs, multiple users LoS channel Sum rate maximization SCA and Lagrange duality

WPT. The first relevant work [215] studied a UAV-mounted maximize the minimum achievable rate among the users with energy transmitter broadcasting wireless energy to charge two their harvested energy constraints, whereas in [224], the max- MTDs. The results showed that when the distance between min harvested energy among users was considered subject the two MTDs is very small, the Pareto boundary of the to users’ achievable rate constraints. An overview on UAV- energy region can be achieved by fixing the UAV’s horizontal enabled SWIPT was presented in [226] by considering the position. In contrast, when the distance between the two context of IoT works for emergency communications. MTDs is large, to achieve the Pareto boundary of the energy On the other hand, different from terrestrial BSs which are region, the UAV should fly and hover between the two MTDs. connected to the power grid, UAVs are typically powered by Subsequently, this work was extended to a more general setup on-board batteries with limited energy storage capacity, which with multiple MTDs in [216] where the authors studied two thus drastically limits their endurance. To overcome the short different objective functions, namely the maximization of the flight time of UAV, several techniques have been proposed, sum energy harvested by all the MTDs and the max-min which can be classified into two categories: external-powered energy harvested among MTDs. Different from the above UAV and mechanical dynamic optimization. works, which obtained locally optimal solutions, a globally 1) External-powered UAV: For external-powered UAVs, optimal one-dimension UAV trajectory design was developed depending on the external sources used for powering the UAV, with the aim of maximizing the minimum energy received by they can be further divided into two approaches, namely solar- either one of the MTDs [217]. powered and laser-powered UAVs. For solar-powered UAVs, a proper size solar panel is installed on the UAV, which first In addition, UAV-aided wireless powered communication harvests solar energy and then converts it to electrical energy network (WPCN) is drawing considerably attention in the re- enabling the UAV to fly and communicate [227]–[229]. Solar- search community [8]. Specifically, receiver devices in WPCN powered UAVs were experimentally verified in [227], [228]. first harvest energy from the signals sent by the UAV in the The results revealed that by carrying a solar panel, the UAV downlink, and then utilize the harvested energy to transmit can continuously fly for more than 24 hours, which renders information to the UAV in the uplink. Depending on the this approach appealing in practice. However, the amount of configuration of the UAVs, the UAV-aided WPCN can be harvested solar energy depends heavily on the flight altitude, classified into two categories. The first is the integrated UAV- weather conditions, and temperature [230], [231]. In particular, aided WPCN, where the energy transmitter and information it was shown in [231] that there exists a tradeoff between the receiver are carried by one UAV [218]–[221]. The other is harvested solar energy and the quality-of-service (QoS) for the separated UAV-aided WPCN, where the energy transmitter the ground users. Specifically, if the UAV operates above the and information receiver are installed on two different UAVs clouds, it can harvest more solar energy but incurs high path [222], [223]. Specifically, the system energy maximization loss between the UAV and the ground users due to the high problem and the max-min throughput optimization problem flight altitude. If the UAV flies below the clouds, less solar in UAV-enabled WPCNs were studied in [218] and [219], energy is available since the solar energy is mainly absorbed respectively. Then, [220] extends the work in [219] to a two- by the clouds, while the quality of the communication link user interference channel for WPCN employing two UAVs. between the UAV and the ground users is improved. Instead of In [221], the multi-agent deep reinforcement learning based exploiting solar energy, a photovoltaic receiver can be installed algorithm was proposed to tackle the max-min optimization on the UAV to harvest power from a laser power beacon. problem in a general multi-UAV enabled WPCN. Further- Compared to solar power, the laser-beam power supply is more more, UAV-enabled simultaneous wireless information and stable and can deliver much more energy to the receiver with power transfer (SWIPT) was proposed recently, where the a narrowly focused energy beam [232]–[235]. In particular, UAV is deployed as an aerial BS to simultaneously send a laser-powered UAV wireless communication system was information and energy to the receivers on the ground [224]– studied in [232], where a laser transmitter sends laser beams [226]. Specifically, the UAV’s trajectory, transmit power, and to charge a fixed-wing UAV in flight, and the UAV uses the users’ power splitting ratios were jointly optimized to the harvested laser energy to communicate with a ground 17 station. In [233], the weighted information transmission effi- handful of works have considered energy-efficient UAV de- ciency and the laser power transmission efficiency of a rotary- sign for different applications such as UAV-assisted mobile wing UAV enabled relay system were maximized subject to edge computing and UAV-assisted backscatter communication the information/energy-causality constraints, the power budget [124], [158], [244], [245]. constraints and the UAVs mobility constraints. It was shown However, current research results only focus on a single that the UAV’s trajectory is highly related to the laser wave- UAV or few UAVs, while energy-efficient design for a large length and the weather condition. number of collaborative UAVs is still a key challenge due 2) Mechanical dynamic optimization: For mechanical dy- to the excessive data and control information exchange be- namic optimization, the research was mainly focused on how tween them needed for efficient cooperation. To tackle this to minimize the UAV’s communication energy and propulsion challenge, some heuristic methods can be applied to the UAV energy consumption [158], [236]–[243]. The UAV’s com- trajectory optimization, such as particle swarm optimization munication energy consumption comprises the circuit power (PSO) as well as its several evolutionary paradigms [246]– consumption as well as the transmit power consumption. Some [249], derivative-free stochastic optimization and Hamiltonian relevant works optimized the deployment of the UAV to min- optimization [250]. Whereas those algorithms are generally imize the transmit power while achieving the different design suboptimal and are not guaranteed to obtain the globally objectives such as the users’ QoS, communication coverage optimal solution, the genetic algorithm (GA) is a useful tool to range, and delay [238]–[240]. However, accurately modelling find optimal solutions for some optimization objectives such the propulsion energy consumption is generally difficult since as energy consumption [248], [251]. However, it leads to a it crucially depends on many practical factors, including the relatively high computational complexity and associated delay UAV velocity, the UAV acceleration, etc. Accordingly, in in practice, especially when the number of UAVs is large. [236], a theoretical propulsion energy consumption model for Recently, game theory-based UAV control methods, e.g. mean- fixed-wing UAVs was derived, which is given by field game, have been exploited as efficient methods to model the interactions among a large number of UAVs [252]–[255].  T  2 (a v) For example, the mean-field approximation was leveraged in c kak − 2 P = c kvk3 + 2 1 + kvk  + maT v , (12) [254] to obtain the near optimal trajectory in the context of 1  2  kvk g UAV-to-device underlaid cellular networks and the approxima-

tion becomes more accurate as the number of involved UAVs where c1 and c2 are constants which are related to the air becomes larger. Furthermore, the joint trajectory design and density, drag coefficient, and wing area; g and m represent the radio resource management problem for cellular Internet of gravitational acceleration and the aircraft’s mass, respectively; UAVs has been studied in [255] by leveraging a cooperative v and a denote the UAV’s velocity and acceleration vectors, sense-and-send protocol. respectively. It is observed from (12) that when the UAV velocity is equal to 0, the UAV’s propulsion energy tends C. Future Research to infinity, which indicates that the UAV trajectory should 1) Air-ground Energy Consumption Tradeoff: For many be carefully designed to avoid excessive energy consumption practical data collection applications in mMTC networks, a and improve endurance. However, for communication service large amount of devices may be distributed in a wide area provisioning, it is desirable to deploy the UAV hovering while each of them only has small bursts of data to transmit. above the ground users [21], [241]. As a result, there is a In this case, having UAVs getting close to each of them to tradeoff between minimizing the UAV’s energy consumption collect data may lead to high propulsion energy consumption. and maximizing the communication throughput. This thus leads to an energy consumption tradeoff between To address this tradeoff, a new metric, namely the energy UAVs and ground devices [25]. To address this issue, inter- efficiency defined as the ratio of the throughput and the UAV UAV cooperation among UAV swarms is a promising solution energy consumption, has been proposed and widely adopted to where a UAV cluster head can be dynamically selected and strike a balance between the above two important design goals other UAVs first collect data in a small area and then transmit [8]. Different from the notion of energy efficiency adopted to the UAV cluster for further processing. in traditional cellular networks, where the energy consump- 2) Access Protocols and Security: For the case of large- tion is related to the transceiver’s circuit power consumption scale deployment of UAVs in heterogenous applications, the and transmit power, energy-efficient designs for UAVs are communication activities of UAVs to the same ground BS also subject to the UAV’s dynamic and kinematic/mobility can be highly random, which may call for random access constraints, which makes the design problems highly non- protocols. Although the grant-based random access protocol convex and much more challenging. For example, the UAVs’ is simpler in practical implementation, as the number of trajectories are continuous and need to satisfy a certain UAVs grows quickly, it would result in a high probability velocity, acceleration and turning angle constraints. Some of access failures due to collision as well as high signaling commonly used methods to solve the resulting optimization overhead arising from handshakes. As a consequence, grant- problems including successive convex approximation (SCA), free random access or unsourced random access protocols Majorization-Minimization (MM) method, block coordinate with properly designed UAV grouping methods help achieve descent method (BCD), alternative direction multiplier method efficient UAV swarm communications [180]. In [256], a semi- (ADMM), penalty based method, machine learning, etc. A grant-free transmission based on NOMA was proposed to offer 18

monitoring UAV

adversary UAV

(a) Sensing for UAV (b) UAV for sensing

Fig. 6. Two paradigms of sensing in UAV networks: (a) Sensing for UAV; (b) UAV for sensing. more refined admission control while minimizing the system and UAV for sensing. In the former case, sensing technolo- signalling overhead. However, a fair and complete comparison gies are utilized to support safe UAV flight and low-altitude between the grant-free random access against NOMA for 3D airspace monitoring and traffic management. By contrast, wireless networks with UAVs remains still an open problem for the paradigm of UAV for sensing, dedicated UAVs are that merits further investigation [257]. Moreover, security and dispatched as aerial flying platforms to provide sensing support privacy are other important issues in safeguarding wireless from the sky. In addition, a comparison of existing works is communication with massive UAV and ground devices [158], shown in Table VIII. [258]–[260], [260]–[263]. A. Sensing for UAV V. RADIO-BASED SENSING Two typical use case scenarios of sensing for UAV are Besides the requirement of high-performance wireless com- sense-and-avoid (SAA) and adversary UAV detection, track- munications, the ability to support effective and efficient sens- ing, and classification. ing is also essential for realizing the vision of integrating UAVs SAA plays an indispensable role to ensure safe UAV flight, into 5G-and-beyond networks. Commercial UAVs nowadays especially for autonomous or semi-autonomous UAVs [264]. are already equipped with a multitude of sensors of various Different from manned aircrafts, UAVs do not have a pilot types, such as inertial measurement unit (IMU), accelerome- onboard, and thus they heavily rely on their own sensed infor- ters, tilt sensors, and current sensors. Such embedded sensors mation for swift response for collision and/or obstacle avoid- provide important real-time information for ensuring safe ance. Note that even for UAVs under real-time remote control, UAV operation, such as the UAV’s position and orientation SAA is important for avoiding catastrophic events, considering estimates, direction and flight path maintaining, and power the long round-trip radio propagation delays and the extra consumption management. response time of ground pilots. To achieve fast response even On the other hand, for future wireless networks, where in highly dynamic environments, SAA typically relies on the UAVs of large-scale deployment will be seamlessly integrated sensed information of on-board sensors, though ground based into terrestrial communication systems, sensing merely relying SAA (GBSAA) is also possible. Note that besides sensing on those on-board embedded sensors will be insufficient. for collision/obstacle avoidance, in many applications such Instead, a combination of both UAV embedded sensing and as UAV-enabled fertilizer or pesticide spraying, sensing may infrastructure-based sensing is needed to achieve high sens- also be necessary for constant-altitude maintenance so as to ing performance, in terms of response time, sensing range, ensure uniform spraying [268], which is a challenging task in coverage, reliability, accuracy, and efficiency. In this section, tough terrains. While many contemporary commercial UAVs we focus on radio-based sensing, where the detection and are already enabled with SAA capabilities, most of them are parameter estimation of the targets of interest are based on vision- or light-based2, which makers them vulnerable to poor the radio signals echoed/scattered by the targets. Compared environmental conditions. with alternative sensing technologies such as acoustic-based, Another important use case of sensing for UAV arises from vision-based, and light-based sensing, radio sensing is less the imperative need for the detection, tracking, and classifica- vulnerable to poor environmental conditions (such as noisy tion of potentially illegitimate and hazardous UAVs. Besides background or dark areas) and can typically support a larger their benign usage, UAVs could also be misused, either inten- sensing range. Besides, the fact that both radio sensing and tionally or unintentionally, which may jeopardize public safety wireless communications utilize radio signals to accomplish and/or threaten privacy. Under such circumstances, UAVs their tasks makes it necessary to study both systems jointly. As illustrated in Fig. 6, radio-based UAV sensing can be 2https://www.dronezon.com/learn-about-drones-quadcopters/ptop-drones- classified into two main paradigms, namely sensing for UAV with-obstacle-detection-collision-avoidance-sensors-explained/ 19

TABLE VIII COMPARISONOFEXISTINGWORKSONRADIO-BASED UAV SENSING.

Reference Main contributions [264] Overview of sense and avoid technologies for unmanned aircraft systems [265] Key techniques of detection, tracking and interdiction of amateur UAVs Sensing for UAV [266] Classification of micro-drones into loaded or unloaded vehicles using multi- static radar [267] Classifying UAVs versus other flying objects such as birds, by extracting the features of the tracks generated by radar [268] Overview for applications of UAV-based sensor platforms and experiments for imaging radar [269] UAV-based remote sensing for field-based crop phenotyping UAV for Sensing [270] UAV-based synthetic aperture radar (SAR) to enhance the radar aperture for high angular resolution [271] Multiple flying UAVs with optimized trajectories to localize moving RF sources [272] Dynamic and reconfigurable aerial radar network composed of UAVs to detect and track other unauthorized/malicious UAVs are usually non-cooperative or even deceptive, which renders mobility of UAVs also introduces a new design degree of applying active detection and localization methods infeasible freedom (DoF) for sensing performance optimization, via 3D [273], but rather requires reliance on radar sensing based on sensor trajectory optimization. This is particularly appealing passively echoed/scaterred signals. While radar sensing for for target tracking, where the UAV locations can be dynami- aircraft detection has long been used in military, its application cally adjusted to best track the target. Therefore, UAV-based for UAVs is facing new challenges. For instance, different from sensing has a wide range of potential applications, such as manned aircrafts, UAVs usually have much smaller radar cross law enforcement, precision agriculture, 3D environment map section (RCS), which makes them more difficult to detect. construction, search and rescue, and military operations. Besides, rotary-wing UAVs are capable of flying at low speed Due to the aforementioned advantages, UAV-based sensing or hovering, which makes them difficult to separate from has received growing interest recently. For instance, various stationary clutter background or other natural flying objects applications of UAV-based sensor platforms were outlined in such as birds. [268], together with a discussion of experiments with an imag- Radar sensing for UAV networks has received significant ing radar. In [269], the authors surveyed UAV-based remote research attention recently. For example, the key techniques sensing for field-based crop phenotyping. In [270], UAV-based of detection, tracking, and interdiction of amateur drones are synthetic aperture radar (SAR) was used to enhance the radar discussed in [265]. Beyond UAV detection, UAV classification aperture for high angular resolution. Besides, multiple flying is also important for safety and privacy protection in the UAVs were utilized in [271] to localize moving RF sources, future internet-of-drones (IoD) era. For example, in [266], the and the UAV trajectory was optimized to minimize a bound authors studied the classification of micro-drones into loaded on the achievable localization error. UAV-aided air quality or unloaded vehicles using multi-static radar. In [267], the sensing was discussed in [274]. In [272], the authors proposed authors studied the problem of classifying UAVs versus other a dynamic and reconfigurable aerial radar network composed flying objects such as birds and manned aircrafts, by extracting of UAVs to detect and track other unauthorized/malicious the features of the tracks generated by radar, where a track is UAVs. It was demonstrated that by exploiting the new DoFs defined as a series of plots associated with each target observed offered by UAV trajectory optimization, the dynamic aerial in consecutive radar scans. radar network is able to improve the tracking performance over a conventional terrestrial radar network with fixed deployment. B. UAV for Sensing C. System Model and Promising UAV Sensing Technologies Another promising paradigm of UAV sensing is to utilize N UAVs as aerial nodes to provide wireless sensing support from For a basic UAV-based radar system with T transmit an- N the sky, which we refer to as UAV for sensing. Compared with tennas and R receive antennas, the input-output relationship conventional ground sensing, UAV-based sensing has several can be modelled as K advantages. First, thanks to its elevated altitude and reduced X (R) (R) T (T ) (T ) signal blockage, UAV-based sensing typically has a wider y(t) = αka(θk , φk )b (θk , φk )x(t − τk) + n(t), field of view (FoV) compared to ground sensors. Besides, k=1 (13) the highly controllable 3D UAV mobility makes it possible to flexibly deploy UAV sensors to hard-to-reach areas, such where K denotes the total number of targets, y(t) ∈ CNR×1 as poisonous or hazardous locations. Furthermore, the high and x(t) ∈ CNT ×1 represent the received and transmitted 20

signal waves, respectively, αk is the complex-valued coeffi- is related to the target distance rk and radial velocity vk as 2(rk+vkt) cient from the transmitter to the receiver associated with the τk = c , where c is the speed of light. kth target, θk and φk are the elevation and azimuth angles The technologies for radar sensing in general have been of the kth target, respectively, with the superscripts (·)(R) advanced tremendously during the past decades. In particular, and (·)(T ) denoting the angle-of-arrival (AoA) and angle-of- it has been shown that MIMO radar is a powerful approach to departure (AoD), respectively. Note that different from ground enhance the radar performance. Specifically, for an NR × NT sensing where the azimuth angle is usually of primary interest, MIMO radar with orthogonal waveforms transmitted from the for UAV-based sensing, both the azimuth and elevation angles NT transmit antennas, we may achieve the same effective are important due to the elevated UAV position. Furthermore, angular resolution as a virtual phased array radar with NT NR a(·) and b(·) denote the receive and transmit array responses, elements [278]–[280]. This thus lays a strong foundation respectively, τk is the radio propagation delay for signals for cellular sensing to achieve super resolution with massive reflected/scattered by the kth target, and n(t) ∈ CNR×1 MIMO in 5G-and-beyond networks. Besides, full-dimensional denotes the receiver noise. Note that in (13), we have only MIMO radar also provides angular resolution in both the included the signal components reflected/scattered by the K azimuth and elevation domains, which is essential for UAV targets of interest. This may be a valid model for UAV sensing sensing. in wide-open areas or after proper clutter subtraction, when the Another promising technology for UAV sensing is mmWave signal components originating from background clutter have sensing, which offers two promising advantages. First, com- been properly eliminated. The radar performance is critically pared to sub-6 GHz systems, physical objects are electri- affected by the radar waveform x(t), which is usually designed cally larger in mmWave systems because of their shorter to have good autocorrelation properties. One popular wave- wavelength. Thus, small objects like micro-UAVs that might form is the linear frequency-modulated continuous waveform be invisible for microwave systems become more visible (FMCW) [275], where multiple chirp signals are transmitted and more easily detectable for mmWave systems [281]. To with the frequency linearly increasing with time. Different illustrate this fact, the authors in [281] used the radar cross- from wireless communication systems, the transmitted radar section (RCS) as a performance measure to compare the waveform x(t) is known at both the transmitter and receiver. delectability of small objects with different frequencies. It was Therefore, based on the knowledge of the transmitted and shown that a small drone illuminated by mmWave radar (60 received waveforms x(t) and y(t), the two fundamental prob- GHz) has a RCS about 30 dB higher than that by a 2 GHz lems in radar sensing are: microwave radar. Second, mmWave systems usually have a Radar detection: Detect the presence/absence of target k, large bandwidth, which leads to high time and range resolu- i.e., αk = 0 for hypothesis H0 and αk 6= 0 for hypothesis tion. In particular, it is known that the radar range resolution H1. In this case, the sensing performance can be measured in ∆ is given by ∆ = c/(2B), where c is the speed of light terms of the probability of correct detection/misdetection and and B is the system bandwidth. This thus makes mmWave the probability of false alarm. systems a promising technology for high-performance radar Radar estimation: If target k is present, estimate the key detection and estimation. Therefore, mmWave UAV detection (T ) (T ) (R) (R) parameters θk , φk , θk , φk , and τk, based on which the has received significant research interest, as reviewed in [281]. target state, such as location and moving velocity, can be further determined. Typical performance metrics for radar D. Joint UAV Communication and Sensing estimation include the accuracy or mean squared error (MSE) Traditionally, wireless communication and radar sensing of the estimation. For unbiased estimators, the classic Cramer- were considered to be two independent systems that are fully Rao lower bound (CRLB) provides the theoretical performance separated in spectrum. With the continuous spectrum expan- bound on the variance of the estimate [276]. In [277], the sion of cellular communication systems (from sub-6 GHz to authors proposed a new performance metric for radar esti- mmWave or even TeraHertz) and the need for developing mation that is similar to communication rate, called radar more intelligent wireless networks, there has been a significant estimation rate. Specifically, the target is treated as a passive research interest recently in joint communication and sensing node with some entropy about its own state that unwillingly (JCAS) [47]–[50]. On the one hand, JCAS facilitates the communicates with the radar receiver, and radar estimation design of a highly flexible and efficient systems utilizing the rate corresponds to how much additional information is gained spectrum allocated for both purposes, which thus offers a new about the target’s state after the radar estimation process. solution approach for resolving the spectrum scarcity. On the Other sensible radar performance metrics include the number other hand, JCAS enables the sharing of wireless infrastructure of resolvable targets, response time, and time and range and RF hardware, so as to build compact and light-weight resolution. wireless equipment with both communication and sensing Note that (13) is applicable for both bi-static and mono- capabilities. Note that this feature is particularly appealing for static radars. In the former, the radar transmitter and receiver UAV networks, considering the severe SWaP and endurance are geographically separated, whereas for the latter, they are constraints of UAVs, together with the significant role that collocated. For mono-static radar, (13) can be further simpli- radar sensing would play towards the integration of UAV into (R) (T ) (R) (T ) fied since we have NT = NR, θk = θk , φk = φk , and 5G-and-beyond cellular networks. b(·) = a(·). Besides, τk corresponds to the round-trip delay The preliminary level of JCAS is communication-radar between the radar transmitter/receiver and the kth target, which coexistence [282], where communication and radar sensing 21 are still designed as two separate systems with their own light-weight, and energy-efficient sensing devices are required respective waveforms and transmitter/receiver designs, but for UAV-based sensing than for conventional ground-based their mutual interference is properly controlled via resource sensing. Besides, the limited endurance of UAVs makes it allocation. This usually corresponds to the scenario when quite challenging to build an aerial sensing network merely radar and communication share neither the transmitter nor the relying on UAVs that is available all day. In this regard, future receiver. For example, a low-altitude UAV that continuously sensing networks are expected to be highly heterogeneous, broadcasts radar waveforms for SAA needs to take into and may utilize complementary ground, aerial, and even space account the potential interference caused to the communication sensing platforms to maximize sensing performance, which between nearby BSs and ground users, and vice versa. deserves in-depth investigations. On the other hand, while A more advanced approach for JCAS is known as dual JCAS has been studied extensively for terrestrial systems, e.g., function radar-communications (DFRC) [283], where both for automotive applications, its application in UAV networks the radar and communication functions are implemented in is still in an early stage. In particular, with elevated UAV the same device. DFRC is ideally suited for application in positions, the coverage requirement for both communication UAV networks. For example, a DFRC-enabled cellular BS and sensing extends from the traditional 2D plane to the 3D may simultaneously communicate with the ground users it airspace, for which cost-effective networking architectures and serves and monitor the low-altitude airspace to detect ad- signal transmission/reception techniques need to be developed. versary UAVs. DFRC can be loosely classified as radar- Besides, when UAVs are utilized as aerial communication and centric, communication-centric, and integrated design. For sensing platforms, the new DoFs offered by UAV trajectory radar-centric DFRC, the transmitted waveform x(t) in (13) design should be exploited for improving the sensing and is radar-oriented, such as the traditional FMCW, but with communication performance. To resolve the SWaP problem additional information-bearing symbols embedded for com- of UAVs, energy-efficient designs for JCAS in UAV networks, munication, e.g., via phase or frequency modulation of the which take into account the UAV propulsion energy as well FMCW pulse [284]. For MIMO radar with transmit beam- as the communication and sensing energy, deserve further forming, the information-bearing symbols may also be em- investigation. Another promising research direction is to ex- bedded into the radar beam pattern, e.g., by amplitude mod- ploit the sensed information, such as the 3D radio propagation ulation of the radar side-lobe levels [285]. While power- environment, to enhance UAV communications [292]. efficient and simple to implement, such radar-centric DFRC can only support very low data rates. On the other hand, VI.AIINTEGRATION for communication-centric DFRC, the system is primarily designed for wireless communications, with the additional In addition to supporting communication and sensing ser- radar functionality. For example, there has been significant vices, the integration of AI is expected to be another important interest in utilizing the standard orthogonal frequency-division feature of future cellular networks, towards the vision of multiplexing (OFDM) communication waveforms for radar network intelligence [51]. Generally speaking, the interplay sensing [286]–[288]. By optimizing the various OFDM pa- between AI and wireless networks has enabled two emerging rameters, such as the sub-carrier spacing and power alloca- paradigms, namely, AI-empowered wireless communications tion, different trade-offs between communication and radar (see, e.g., [51], [52]) and edge intelligence (see e.g., [293], sensing performance can be achieved. However, for high mo- [294]). In the former paradigm, AI and machine learning bility scenarios like UAV networks, the sensing performance techniques are utilized as a new data-driven mathematical (such as the unambiguous range) with OFDM waveforms tool (in contrast to conventional model-driven methods) for is vulnerable to sub-carrier misalignment [289]. Lastly, for optimization of wireless systems to enhance the communi- integrated radar-communication design, the DFRC waveform cation performance. In the latter paradigm, AI and mobile might be derived from the scratch without restriction to edge computing (MEC) capabilities [295], [296] are incor- radar- or communication-centric designs, which makes further porated into BSs and APs at the network edge, for enabling performance improvements possible. For example, in [290], various intelligent applications (such as autonomous driving the weighted combination of the radar and communication and industrial automation) with extensive communication and waveforms was used as the transmitted waveform of a DFRC computation requirements [293], [294]. As compared to the MIMO transmitter, and the resulting transmit covariance ma- conventional cloud and on-device intelligence, edge intelli- trix was optimized to preserve the desired radar beampattern, gence can significantly reduce the end-to-end latency and while guaranteeing a minimum required SINR requirement at minimize the traffic loads to the core network. each communication user. Beamforming for JCAS with analog Following the above two paradigms, AI is expected to or hybrid analog/digital antenna architectures was studied in also play an important role for 5G-and-beyond UAV com- [291] and [47], respecitvely. munication networks in two aspects. On the one hand, for highly dynamic UAV-enabled 3D networks, AI and machine learning can serve as promising alternative mathematical tools E. Future Research to solve, e.g., joint UAV trajectory and communication design The research on sensing in UAV networks is still in its in- problems, where the conventional model-driven optimization fancy, with many new opportunities and challenges ahead. For may not work well due to the difficulty in obtaining accurate example, due to the severe SWaP constraint, more compact, network state information. On the other hand, the integration 22 of UAVs in edge intelligence not only facilitates emerging and cellular-connected UAVs [308], [309], when the UAVs applications (such as drone VR and drone swarms), but also in- serve as BSs and users, respectively, as elaborated in the troduces new challenges in effectively handling computation- following. Besides, a comparison of existing works in this intensive and latency-critical AI tasks from the sky. This calls field is provided in Table IX. for the joint design of the UAVs’ mobility/trajectory control 1) UAVs as BSs: First, consider the case where one sin- together with the communication and computation resource gle UAV serves as an aerial BS for UAV-assisted wireless allocation, which, however, may be particularly difficult as communications and needs to simultaneously serve multiple UAVs may act as aerial users or aerial edge servers or both. users on the ground. In this case, the UAV needs to find the Notice that although there have been prior works [297], optimal flight trajectory maximizing the performance, while [298] that reviewed the design of UAV communication net- taking into account the wireless channel conditions of all works based on machine learning, there still lacks an overview users. However, as obstacles (e.g., buildings or trees) are of the integration of AI in UAV communication networks non-uniformly distributed over space, the UAV may possess from the above two aspects. In this section, we first review LoS and non-LoS (NLoS) links with different ground users machine learning methods for UAV trajectory and commu- at different locations. In practice, as the obstacles’ locations nications design, then discuss computation offloading design may not be known prior to the optimization, the LoS/NLoS for UAVs with MEC, and finally present distributed edge property of the wireless channels becomes uncertain, thus machine learning with UAVs, followed by some open research making it infeasible to formulate the UAV trajectory design problems. as an explicit optimization problem. Although prior works as- sumed probabilistic LoS channel models and accordingly used A. Machine Learning for UAV Trajectory and Communication average channel power gains in their problem formulations [8], Design such designs would lead to highly compromised solutions. In contrast, RL is an efficient model-free method, which can learn The joint design of the UAVs’ movements over time the environment on-the-fly based on the past experiences, and (e.g., deployment locations and flight trajectories) and the automatically generate optimized solutions. For instance, the communication resource allocation is crucial for optimiz- authors in [301] assumed that one UAV communicates with ing the performance of 5G-and-beyond UAV communication multiple ground users over orthogonal channels, where the networks. Conventionally, such joint design is implemented UAV aims to maximize the communication sum rate for all in an offline manner based on model-driven optimization users by optimizing the trajectory. By applying Q-learning approaches, where the network state information (such as (a model-free RL technique), the UAV can autonomously the locations of the communication nodes, the CSI, and the learn its optimal trajectory for sum-rate maximization, without service request information) is assumed to be (perfectly or explicitly having the environment information. Furthermore, partially) known prior to the optimization. In this case, the the authors in [302] studied the sum-rate maximization over joint design can be formulated as a deterministic optimization a finite duration in a UAV-enabled uplink NOMA system for problem that is generally solvable via convex and non-convex data collection, where the ground users may move dynamically optimization techniques. Such offline designs, however, may over time. By assuming that the UAV only causally knows not work well in practice, due to the time- and spatial-varying the users’ (moving) locations and CSI, a Q-learning-based traffic demands, the user mobility, and the complicated channel RL approach was proposed to enable the UAV to optimize propagation environments introduced by the UAVs. To tackle its flying path in an online manner, where expert knowledge this challenge, researchers have shifted from conventional of well-established wireless channel models was exploited to model-driven approaches to alternative data-driven approaches initialize the Q-table values. by exploiting emerging machine learning techniques. Besides the model-free design without any LoS/NLoS infor- In general, machine learning can be classified into three mation in [301], [302], environment maps [303], [323], which categories, namely supervised learning, unsupervised learning, contain the LoS/NLoS channel information over space, can and RL. RL is particularly useful for the joint UAV movement be further exploited for more efficient trajectory design. For and communication design, and thus will be the main focus example, the authors in [303] considered the UAV-enabled data of this subsection.3 To be specific, RL optimizes the actions collection over a finite flight duration subject to obstacle avoid- of one or more agents in an environment to maximize the ance constraints, for which a new end-to-end RL approach was cumulative reward over a certain time horizon [299]. Sup- proposed to efficiently control the UAV trajectory for enhanc- posing that UAVs are the agents of interest, then RL can ing the communication performance. For this setup, a double be efficiently utilized in UAV communication networks for deep Q-network (DDQN) with combined experience replay rapidly adapting to the dynamic environment [300], by prop- was trained to learn the UAV control policy. By exploiting a erly choosing the UAVs’ actions (e.g., deployment/trajectory multi-layer map of the environment, the DDQN is applicable design and resource allocations) and the reward functions (e.g., for general scenarios when the system and channel parameters communication rate). In the literature, there are two lines (such as the number of ground users, their locations, and the of research that exploit RL for optimizing the UAVs’ op- maximum flight duration) can vary. eration in UAV-assisted communication systems [300]–[307] Next, when multiple UAVs are deployed in a network for 3Please refer to [298] for reviews of using other machine learning methods serving multiple users on the ground, it becomes important in the design of UAV communication networks. for the UAVs to jointly design their trajectories and com- 23

TABLE IX COMPARISONOFEXISTINGWORKSON AI-INTEGRATION IN UAV COMMUNICATION NETWORKS.

Reference Main contributions [301], [302] Q-learning-based trajectory design for one UAV communicating with multiple ground users [303] Joint exploitation of double deep Q-network (DDQN) and environment maps Machine Learning for UAVs for trajectory control in UAV-enabled data collection [304] Three-step machine learning based approach for optimizing cell partition, 3D deployments, and dynamic movements in multi-UAV-enabled networks [305] Decentralized deep RL for multiple UAV-BSs’ trajectories design to minimize the AoI [308] DDQN for simultaneous navigation and radio mapping [309] Deep Q-learning for optimizing the height of UAV to balance between desirable BS’s signals and other BSs’ inteference [310] Joint trajectory and TDMA-based transmission design for one UAV-edge-device to offload computation tasks to multiple BSs for parallel execution [311]–[313] Joint communication, computation, and trajectory optimization for UAV-edge- Computation Offloading server to maximize the computation performance of ground devices [314] UAV-relay-assisted MEC [315], [316] UAV-enabled wireless powered MEC [317], [318] Joint trajectory and resource allocation design for multi-UAV-enabled MEC networks [319] Wireless resource allocation for UAV-enabled federated edge learning [320] Federated edge learning for UAV swarms Edge Machine Learning [321] UAVs as edge servers for federated learning [322] UAV-enabled federated learning for Internet of vehicles munications to collaboratively enhance the network coverage as high buildings. To tackle this challenge, using machine and throughput performance. For instance, in [304], a three- learning based designs for refining the UAV trajectories are step machine learning based approach was proposed to max- emerging to maximize the connectivity probability during a imize the satisfaction of the users, where the cell partition, task. For instance, in the case with one single UAV, the authors the 3D deployments of multiple UAVs, and their dynamic in [308] proposed a deep RL-based solution approach, namely movements are sequentially optimized via K-means and two a dueling double deep Q-network (dueling DDQN), to design Q-learning-based algorithms, respectively. Furthermore, de- the navigation/trajectory for optimally balancing between the centralized deep RL based trajectory designs were developed UAV mission completion time versus the expected duration of to ensure coverage and user fairness while maximizing the communication outage. More specifically, this design exploits energy efficiency of UAVs [305] and to minimize the age- the dual use of the UAV’s signal measurements not only for of-information (AoI) when the UAVs execute sensing tasks training the DQN, but also for creating a radio map to predict through cooperative sensing and transmission [307]. Moreover, the outage probabilities at different locations, thus leading to the authors in [306] considered the scenario of UAV-enabled an interesting framework termed simultaneous navigation and caching, where a distributed algorithm based on the machine radio mapping (SNARM) that greatly improves the learning learning framework of liquid state machine (LSM) was devel- performance. oped to optimize the content caching and resource allocation decisions at the UAVs. 2) UAVs as Users: In cellular-connected UAV scenarios, UAVs act as aerial users in cellular networks to perform tasks Besides the signal blockage due to obstacles in the commu- like packet delivery and aerial inspection. In this scenario, how nication path, the co-channel interference among nearby BSs to maintain the cellular connection while addressing the task is is another issue that may affect the system performance. Recall the main concern. Normally, conventional designs optimize the that the signal path loss for UAV-ground channels is altitude- UAV trajectories based on distance-dependent channel models dependent, i.e., when the UAV stays at a higher altitude, under a fixed path loss exponent. These works, however, the LoS probability of the UAV-ground channels increases. overlook the complicated air-ground channel propagation en- Exploiting such properties, deep Q-learning was used in [309] vironments, and thus may lead to unexpected communication for optimizing the height of the UAV for maximization of outage at the cellular BSs in coverage holes where the wireless the spectral efficiency, via properly balancing between the channels of the ground BSs are blocked by obstacles such desirable signal and harmful interference. 24

Fig. 7. Computation offloading with UAV: (a) cellular-connected UAV; (b) UAV-assisted MEC system.

B. Computation Offloading with UAVs for AI Tasks same BS to which the tasks were offloaded, thus resulting in the coupling between (uplink) offloading and (downlink) To enable edge intelligence applications with both UAVs downloading. This makes the trajectory design even more and AI integrated, a large volume of data are to be gener- difficult, as the UAV may need to fly back and forth during ated at distributed edge devices (including both UAVs and offloading and downloading. conventional smart sensors and smart phones), which need There have been various attempts to address the aforemen- to be properly processed via sophisticated AI training and tioned problems. For example, the authors in [310] proposed inference algorithms. However, the implementation of these a TDMA-based protocol for computation offloading, where AI tasks is generally data- and computation-intensive, which the UAVs can offload their tasks to different BSs for parallel cannot be handled locally by wireless devices themselves. As execution in orthogonal time slots, in order to fully exploit such, computation task offloading is an appealing solution the computation resources available at the distributed BSs. In to handle AI tasks, which allows the UAV edge servers to particular, the UAV’s mission completion time is minimized offload their AI tasks to MEC servers with high computation by jointly optimizing the UAV trajectory and the computation capabilities, and then download the computation results after offloading scheduling, subject to the UAV’s flight constraints, MEC execution. Fig. 7(a) shows the scenario when a UAV acts and the ground BSs’ (GBSs’) computation capacity con- as an aerial user or edge device in cellular networks, which straints. For illustration, Fig. 8 shows the optimized UAV tra- has certain computation tasks to be executed via offloading jectories for computation offloading for different computation to BSs on the ground. Besides edge devices, UAVs may loads (measured in computation task input bits L [310]). There also carry MEC servers to support the AI implementation are four GBSs and the UAV’s initial and final coordinates are of on-ground devices. As shown in Fig. 7(b), the UAV can (0,0) and (1000 m, 1000 m), respectively. It is observed that help the widely distributed devices on the ground to execute when the computation load is small (i.e., L = 100 Mbits), their computation-intensive AI tasks, especially in emergency the UAV flies in a straight line from the initial to the final situations. Under both scenarios, the joint UAV trajectory and location. When L increases to 200 Mbits, the flight trajectory communication/computation design is crucial. For facilitate deviates from the straight line for more efficient offloading. such joint design, in this line of research the AI tasks are nor- For L = 600 Mbits, the UAV flies back and forth between mally modeled as general computation tasks with certain data different GBSs in order to exploit multiple GBSs’ distributed and computation requirements, as detailed in the following. computation resources more efficiently via time sharing. This 1) UAVs as Edge Devices: First, as shown in Fig. 7(a), phenomenon is significantly different from the UAV trajectory assume that a UAV user needs to fly from an initial to design for communications, where the UAV may successively a final location (e.g., for packet delivery) and has certain visit different ground nodes, instead of flying back and forth computation tasks to be executed during the mission (e.g., for computation offloading because of the inherent distributed processing of sensed data in real time for autonomous flight). computation capacity constraints. During the flight, the UAV needs to offload its computa- 2) UAVs as Edge Servers: In another line of research, tion tasks to ground BSs/MEC servers, and download the the UAVs can be utilized as MEC servers in the sky to computation results from the corresponding BSs after their remotely execute the computation tasks of ground devices, remote execution. In this case, the UAV’s trajectory design as shown in Fig. 7(b), which may correspond to scenarios becomes a more complicated problem due to the distributed when the ground infrastructure is damaged or not available. As computation constraints at the BSs. For instance, a UAV may the UAV’s limited communication and computation resources prefer to partition the computation tasks and offload them to are shared among multiple ground devices, the UAV should different BSs to exploit their cooperative computation gain; carefully design the joint communication and computation and accordingly, the trajectory must be carefully design in scheduling, together with the trajectory design, to maximize order to meet the offloading-execution-downloading require- the computation performance of ground devices in a fair and ments of different BSs. It should be further noted that the efficient manner [311]–[313]. UAV can only download the computation results from the Furthermore, the single-UAV-assisted computation offload- 25

Moreover, researchers also investigated UAV swarm appli- 1000 UAV Trajectory (L = 100 Mbits) GBS 4 cations with stringent computation requirements. Towards this UAV Trajectory (L = 200 Mbits) 900 UAV Trajectory (L = 600 Mbits) end, one efficient solution is to employ high-altitude UAVs Initial and final points 800 GBSs with larger size as edge servers to support the swarm UAVs’ computation [326], while another solution is to use the edge 700 and cloud infrastructures on the ground to support the swarm 600 UAVs’ extensive computation tasks [327].

500 y(m) C. Distributed Edge Machine Learning with UAVs 400 GBS 1 GBS 3 Besides computation offloading towards a central MEC 300 server, edge devices such as UAVs can also collaboratively

200 GBS 2 perform AI tasks using their locally distributed data and

100 computation capabilities. This technique is generally referred to as distributed edge learning. Among other approaches, 0 0 100 200 300 400 500 600 700 800 900 1000 federated edge learning, initially developed by Google [328], is x(m) particularly important due to its advantages in preserving data security and privacy [294], where a group of edge devices Fig. 8. Obtained UAV trajectory for computation offloading for different (UAVs of interest) jointly use their distributed data to train computation loads. common AI or machine learning models without data sharing. As illustrated in Fig. 9(a) and Fig. 9(b), the edge server can either be a BS on the ground (e.g., as with a cellular- ing in Fig. 7(b) has been extended to various other setups. connected UAV) or another UAV in the sky (e.g., as in a UAV For instance, the authors in [314] presented a UAV-relay- swarm). Federated edge learning is implemented in an iterative assisted MEC system by integrating UAVs into the cooperative manner: in each iteration, the UAVs first update their local AI MEC design [324], where the UAV not only acts as the models, and then aggregate at the edge server to update global MEC server for remote task execution, but also serves as a AI models. The above process requires the UAVs and edge relay to help ground devices offload tasks to another GBS. server to frequently exchange their AI model parameters, and In this system, the UAV exploits the joint communication iterate between communication and computation operations. and computation cooperation for computation performance Because of this fact and the UAVs’ 3D mobility, how to jointly enhancement. Moreover , the authors in [315], [316] proposed optimize their multiple UAVs’ trajectories together with the UAV-enabled MEC wireless-powered systems by integrating communication and computation scheduling over time is the the UAV with wireless powered MEC [296], where the UAV key issue to be tackled. serves as energy transmitter, communication transceiver, and While federated edge learning is currently a very hot topic MEC server. In this system, the UAV trajectory is optimized in both the wireless communications and machine learning jointly with the energy/communication/computation resource societies, the research on UAV-enabled edge learning is still allocations for maximization of the computation performance in its infancy. For instance, [329] presented various applica- of the ground devices, subject to a new type of wireless tions of federated learning in UAV wireless communication energy harvesting constraint as the energy consumption for networks. Wireless resource allocation for enabling efficient communication and computation at each device cannot exceed UAV-enabled federated edge learning was studied in [319]. the energy harvested wirelessly from the UAV. In addition, Federated learning for UAV swarms was investigated in security concerns during offloading from ground devices to the [320]. Furthermore, using UAVs as edge servers for federated UAV were addressed in [325], where physical layer security learning was proposed in [321], and the support of Internet is employed to combat against potential eavesdropping from of vehicles was considered in [322]. Despite this research ground attackers. progress, the fundamental performance limits of federated There have also been a handful of works investigating UAV- learning with mobile UAV nodes are still a largely uncharted assisted computation offloading to multiple UAV-mounted area. MEC servers [317], [318]. In this case, the association between the ground devices and the UAVs is an important new issue D. Future Research to be considered, together with UAV trajectory and resource 1) Cloud-Edge-Device Collaboration with UAVs: As UAVs allocation design [317]. Different from conventional designs may perform as both aerial users and service (communi- focusing on communication only, the user association in the cation/computation/AI) providers, future networks need to MEC context needs to properly balance both the communi- be aerial-ground integrated, where computation resources cation and computation loads, thus making it more difficult. are distributed everywhere across heterogeneous aerial and Furthermore, the self-interest is another issue for multi-UAV ground nodes with distinct communication and computation networks, especially when the UAVs may belong to different capabilities (see, e.g., [330]). How to match the time- and service providers [318]. To address this issue, coalition forma- spatial-varying communication/computation/AI demands with tion for UAVs can be investigated by using game theory and distributed communication/computation/data supplies in such RL. highly dynamic 3D networks is a challenging task. 26

Fig. 9. Federated edge learning with UAVs: (a) the case with coordination of ground BS; (b) the case with UAV peers.

2) Over-the-Air Computation From Sky: Over-the-air com- [8] Y. Zeng, Q. Wu, and R. Zhang, “Accessing from the sky: A tutorial putation (AirComp) is an emerging technique to enable on UAV communications for 5G and beyond,” Proc. IEEE, vol. 107, no. 12, pp. 2327–2375, Dec. 2019. functional computation over wireless channels by exploit- [9] M. Mozaffari, W. Saad, M. Bennis, Y.-H. Nam, and M. Debbah, “A ing wireless signal superposition property of multiple access tutorial on UAVs for wireless networks: applications, challenges, and channels [331]–[335]. AirComp leads to a fully-integrated- open problems,” IEEE Commun. Surveys Tuts., vol. 21, no. 3, pp. 2334– communication-computation design, and thus can have abun- 2360, Third quarter 2019. [10] Y. Zeng, I. Guvenc, R. Zhang, G. Geraci, and D. W. Matolak, UAV dant applications in distributed sensing, computation, federated Communications for 5G and Beyond. Wiley Online Library, 2020. learning [336], [337], and UAV swarms. When UAVs are [11] M. Shafi, A. F. Molisch, P. J. Smith, T. Haustein, P. Zhu, P. De Silva, involved in AirComp as either edge devices or servers, how to F. Tufvesson, A. Benjebbour, and G. Wunder, “5G: A tutorial overview of standards, trials, challenges, deployment, and practice,” IEEE J. Sel. exploit their mobility for optimizing the AirComp performance Areas Commun., vol. 35, no. 6, pp. 1201–1221, Jun. 2017. is an interesting problem that is uncharted yet. [12] P. Popovski, K. F. Trillingsgaard, O. Simeone, and G. Durisi, “5G wireless network slicing for eMBB, URLLC, and mMTC: A communication-theoretic view,” IEEE Access, vol. 6, pp. 55 765– VII.CONCLUSIONS 55 779, 2018. In this paper, we provided a comprehensive overview on [13] H. Ji, S. Park, J. Yeo, Y. Kim, J. Lee, and B. Shim, “Ultra-reliable and low-latency communications in 5G downlink: Physical layer aspects,” the integration of UAVs into 5G-and-beyond networks, which IEEE Wireless Commun., vol. 25, no. 3, pp. 124–130, Jun. 2018. is expected to drive UAV verticals and cellular networks to [14] C. She, C. Liu, T. Q. S. Quek, C. Yang, and Y. Li, “Ultra-reliable and a win-win situation by jointly exploiting their high altitude low-latency communications in unmanned aerial vehicle communica- tion systems,” IEEE Trans. on Commun., vol. 67, no. 5, pp. 3768–3781, and flexible 3D movement together with the most advanced May 2019. wireless technologies. In addition to reviewing the state-of-the- [15] C. Bockelmann, N. Pratas, H. Nikopour, K. Au, T. Svensson, C. Ste- art research from both the academic and industrial viewpoints, fanovic, P. Popovski, and A. Dekorsy, “Massive machine-type commu- nications in 5G: Physical and MAC-layer solutions,” IEEE Commun. we focused on the most promising solutions of addressing Mag., vol. 54, no. 9, pp. 59–65, Sep. 2016. the challenges in communication, sensing, and AI, and also [16] Y. Zeng, J. Lyu, and R. Zhang, “Cellular-connected UAV: Potential, exploring new opportunities in system design and implemen- challenges, and promising technologies,” IEEE Wireless Commun., tation of UAV cellular networks. Promising directions and vol. 26, no. 1, pp. 120–127, Feb. 2018. [17] W. Mei, Q. Wu, and R. Zhang, “Cellular-connected UAV: Uplink open problems in 3D wireless networks with UAVs were also association, power control and interference coordination,” IEEE Trans. highlighted for future research. As the research in this frontier Wireless Commun., vol. 18, no. 11, pp. 5380–5393, Nov. 2019. paradigm is new and remains largely unexplored, it is hoped [18] S. Zhang, Y. Zeng, and R. Zhang, “Cellular-enabled UAV communi- cation: a connectivity-constrained trajectory optimization perspective,” that this paper will serve as a timely and useful resource for IEEE Trans. Commun., vol. 67, no. 3, pp. 2580–2604, Mar. 2019. researchers to push forward. [19] Y. Zeng, R. Zhang, and T. J. Lim, “Wireless communications with un- manned aerial vehicles: opportunities and challenges,” IEEE Commun. Mag., vol. 54, no. 5, pp. 36–42, May 2016. REFERENCES [20] Q. Wu, L. Liu, and R. Zhang, “Fundamental tradeoffs in communica- [1] https://www.droneii.com/5g-and-drones-part-1-what-is-5g. tion and trajectory design for UAV-enabled wireless network,” IEEE [2] https://www.marketsandmarkets.com/pdfdownloadNew.asp?id=662. Wireless Commun., vol. 26, no. 1, pp. 36–44, Feb. 2019. [3] https://www.faa.gov/news/speeches/news story.cfm?newsid=22294. [21] Q. Wu, Y. Zeng, and R. Zhang, “Joint trajectory and communication [4] https://qz.com/1596735/firefighters-used-drones-to-tackle-notre-dame-fire/. design for multi-UAV enabled wireless networks,” IEEE Trans. Wire- [5] https://www.geospatialworld.net/blogs/ less Commun., vol. 17, no. 3, pp. 2109–2121, Mar. 2018. how-drones-are-being-used-to-combat-covid-19/. [22] Y. Zeng, R. Zhang, and T. J. Lim, “Throughput maximization for UAV- [6] A. Fotouhi, H. Qiang, M. Ding, M. Hassan, L. G. Giordano, A. Garcia- enabled mobile relaying systems,” IEEE Trans. Commun., vol. 64, Rodriguez, and J. Yuan, “Survey on UAV cellular communications: no. 12, pp. 4983–4996, Dec. 2016. practical aspects, standardization advancements, regulation, and secu- [23] M. T. Islam, A. M. Taha, and S. Akl, “A survey of access management rity challenges,” to appear in IEEE Commun. Surveys Tuts., 2019. arXiv techniques in machine type communications,” IEEE Commun. Mag., preprint arXiv:1809.01752. vol. 52, no. 4, pp. 74–81, Apr. 2014. [7] M. Hassanalian and A. Abdelkefi, “Classifications, applications, and [24] Q. Wu and R. Zhang, “Common throughput maximization in UAV- design challenges of drones: A review,” Progress in Aerospace Sci- enabled OFDMA systems with delay consideration,” IEEE Trans. ences, vol. 91, pp. 99–131, May 2017. Commun., vol. 66, no. 12, pp. 6614–6627, Dec. 2018. 27

[25] D. Yang, Q. Wu, Y. Zeng, and R. Zhang, “Energy tradeoff in ground-to- [54] X. Lin, et al., “The sky is not the limit: LTE for unmanned aerial UAV communication via trajectory design,” IEEE Trans. Veh. Technol., vehicles,” IEEE Commun. Mag., vol. 56, no. 4, pp. 204–210, Apr. vol. 67, no. 7, pp. 6721–6726, Jul. 2018. 2018. [26] J. Lyu, Y. Zeng, and R. Zhang, “Cyclical multiple access in UAV- [55] Https://www.3gpp.org/uas-uav. aided communications: A throughput-delay tradeoff,” IEEE Wireless [56] Https://www.comsoc.org/about/committees/emerging-technologies- Commun. Lett., vol. 5, no. 6, pp. 600–603, Dec. 2016. initiatives/aerial-communications. [27] https://www.qualcomm.com/news/onq/2017/05/03/ [57] Https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9097766. qualcomm-technologies-releases-lte-drone-trial-results. [58] Https://aceti.committees.comsoc.org/initiatives/. [28] https://about.att.com/story/att and intel to test drones on lte [59] E. W. Frew and T. X. Brown, “Airborne communication networks for network.html. small unmanned aircraft systems,” Proc. IEEE, vol. 96, no. 12, pp. [29] https://www.mobileworldlive.com/asia/asia-news/ 2008–2027, Dec. 2008. china-mobile-ericsson-test-drone-on-5g-enabled-network. [60] B. V. D. Bergh, A. Chiumento, and S. Pollin, “LTE in the sky: trading [30] https://www.nokia.com/about-us/news/releases/2016/10/03/f-cell- off propagation benefits with interference costs for aerial nodes,” IEEE technology-from-nokia-bell-labs-revolutionizes-small-cell-deployment- Commun. Mag., vol. 54, no. 5, pp. 44–50, May 2016. by-cutting-wires-costs-and-time/. [61] I. B. Yaliniz and H. Yanikomeroglu, “The new frontier in RAN IEEE Commun. Mag. [31] https://www.fiercewireless.com/wireless/ heterogeneity: multi-tier drone-cells,” , vol. 54, verizon-tests-flying-cell-site-drone-as-part-emergency-preparedness. no. 11, pp. 48–55, Nov. 2016. [62] S. Sekander, H. Tabassum, and E. Hossain, “Multi-tier drone architec- [32] https://www.geekwire.com/2017/facebook-tether-tenna-internet-helicopters/. ture for 5G/B5G cellular networks: challenges, trends, and prospects,” [33] https://smallcells.3g4g.co.uk/2019/03/huawei-skysite-drone-with-5g-base. IEEE Commun. Mag., vol. 56, no. 3, pp. 96–103, Mar. 2018. html. [63] Y. Zeng, J. Lyu, and R. Zhang, “Cellular-connected UAV: potentials, [34] https://www.percevite.org/. challenges and promising technologies,” IEEE Wireless Commun., [35] F. Minucci, E. Vinogradov, and S. Pollin, “Avoiding collisions at any vol. 26, no. 1, pp. 120–127, Feb. 2019. (low) cost: ADS-B like position broadcast for UAVs,” IEEE Access, [64] W. Khawaja, I. Guvenc, D. W. Matolaky, U. C. Fiebigz, and N. Sch- vol. 8, pp. 121 843–121 857, Jul. 2020. neckenberger, “A survey of air-to-ground propagation channel modeling [36] https://www.droc2om.eu/. for unmanned aerial vehicles,” IEEE Commun. Surveys Tuts., vol. 21, [37] H. C. Nguyen, R. Amorim, J. Wigard, I. Z. Kovacs,´ T. B. Sørensen, and no. 3, pp. 2361–2391, Third quarter 2019. P. E. Mogensen, “How to ensure reliable connectivity for aerial vehicles [65] A. A. Khuwaja, Y. Chen, N. Zhao, M.-S. Alouini, and P. Dobbins, “A over cellular networks,” IEEE Access, vol. 6, pp. 12 304–12 317, Mar. survey of channel modeling for UAV communications,” IEEE Commun. 2018. Surveys Tuts., vol. 20, no. 4, pp. 2804–2821, Fourth quarter 2018. [38] https://www.sesarju.eu/projects/secops. [66] E. W. Frew and T. X. Brown, “Airborne communication networks for [39] http://www.absolute-project.eu. small unmanned aircraft systems,” Proc. IEEE, vol. 96, no. 12, pp. [40] S. Chandrasekharan et al., “Designing and implementing future aerial 2008–2027, Dec. 2008. communication networks,” IEEE Commun. Mag., vol. 54, no. 5, pp. [67] I. Bekmezci, O. K. Sahingoz, and S¸. Temel, “Flying ad-hoc networks 26–34, May 2016. (FANETs): a survey,” Ad Hoc Networks, vol. 11, no. 3, pp. 1254–1270, [41] https://gow.epsrc.ukri.org/NGBOViewGrant.aspx?GrantRef=EP/ Jan. 2013. P028764/1. [68] L. Gupta, R. Jain, and G. Vaszkun, “Survey of important issues in UAV [42] G. Deepak, A. Ladas, Y. A. Sambo, H. Pervaiz, C. Politis, and communication networks,” IEEE Commun. Surveys Tuts., vol. 18, no. 2, M. A. Imran, “An overview of post-disaster emergency communication pp. 1123–1152, Second quarter 2016. systems in the future networks,” IEEE Wireless Commun., vol. 26, [69] S. Hayat, E. Yanmaz, and R. Muzaffar, “Survey on unmanned aerial no. 6, pp. 132–139, Dec. 2019. vehicle networks for civil applications: a communications viewpoint,” [43] https://5gdrones.eu/. IEEE Commun. Surveys Tuts., vol. 18, no. 4, pp. 2624–2661, Fourth [44] S. Si-Mohammed, M. Bouaziz, H. Hellaoui, O. Bekkouche, A. Ksen- quarter 2016. tini, T. Taleb, L. Tomaszewski, T. Lutz, G. Srinivasan, T. Jarvet, and [70] H. Shakhatreh, A. Sawalmeh, A. Al-Fuqaha, Z. Dou, E. Almaita, P. Montowtt, “Supporting unmanned aerial vehicle services in 5G I. Khalil, N. S. Othman, A. Khreishah, and M. Guizani, “Unmanned networks: New high-level architecture integrating 5G with U-space,” aerial vehicles: A survey on civil applications and key research chal- IEEE Veh. Technol. Mag., vol. 16, no. 1, pp. 57–65, Mar. 2021. lenges,” IEEE Access, vol. 7, pp. 48 572–48 634, Apr. 2019. [45] https://aerpaw.org/. [71] N. H. Motlagh, T. Taleb, and O. Arouk, “Low-altitude unmanned aerial vehicles-based internet of things services: comprehensive survey and [46] V. Marojevic, I. Guvenc, R. Dutta, M. L. Sichitiu, and B. A. Floyd, future perspectives,” IEEE Internet Things J., vol. 3, no. 6, pp. 899– “Advanced wireless for unmanned aerial systems: 5G standardization, 922, Dec. 2016. research challenges, and AERPAW architecture,” IEEE Veh. Technol. [72] R. Shakeri, M. A. Al-Garadi, A. Badawy, A. Mohamed, T. Khattab, Mag., vol. 15, no. 2, pp. 22–30, Jun. 2020. A. K. Al-Ali, K. A. Harras, and M. Guizani, “Design challenges of [47] F. Liu, C. Masouros, A. P. Petropulu, H. Griffiths, and L. Hanzo, “Joint multi-UAV systems in cyber-physical applications: A comprehensive radar and communication design: Applications, state-of-the-art, and the survey and future directions,” IEEE Commun. Surveys Tuts., vol. 21, IEEE Trans. Commun. road ahead,” , vol. 68, no. 6, pp. 3834–3862, Jun. no. 4, pp. 3340–3385, Fourth quarter 2019. 2020. [73] Q. Wu, G. Y. Li, W. Chen, D. W. K. Ng, and R. Schober, “An overview [48] M. L. Rahman, J. A. Zhang, K. Wu, X. Huang, Y. J. Guo, S. Chen, and of sustainable green 5G networks,” IEEE Wireless Commun., vol. 24, J. Yuan, “Enabling joint communication and radio sensing in mobile no. 4, pp. 72–80, Aug. 2017. networks–a survey,” arXiv preprint arXiv:2006.07559, 2020. [74] S. Zhang, Q. Wu, S. Xu, and G. Y. Li, “Fundamental green tradeoffs: [49] M. L. Rahman, J. A. Zhang, X., Huang, Y. J. Guo, and R. W. Heath, progresses, challenges, and impacts on 5G networks,” IEEE Commun. “Framework for a perceptive mobile network using joint communi- Surveys Tuts., vol. 19, no. 1, pp. 33–56, First Quarter 2017. cation and radar sensing,” IEEE Trans. Aerospace and Electr. Syst., [75] Q. Wu, G. Y. Li, W. Chen, and D. W. K. Ng, “Energy-efficient vol. 56, no. 3, pp. 1926–1941, Jun. 2020. small cell with spectrum-power trading,” IEEE J. Sel. Areas Commun., [50] Z. Feng, Z. Fang, Z. Wei, X. Chen, Z. Quan, and D. Ji, “Joint radar vol. 34, no. 12, pp. 3394–3408, Dec. 2016. and communication: A survey,” China Commun., vol. 17, no. 1, pp. [76] ——, “Energy-efficient D2D overlaying communications with 1–27, Jan. 2020. spectrum-power trading,” IEEE Trans. Wireless Commun., vol. 16, [51] K. B. Letaief, W. Chen, Y. Shi, J. Zhang, and Y. A. Zhang, “The no. 7, pp. 4404–4419, Jul. 2017. roadmap to 6G: AI empowered wireless networks,” IEEE Commun. [77] Y. Huang, Q. Wu, R. Lu, X. Peng, and R. Zhang, “Massive MIMO for Mag., vol. 57, no. 8, pp. 84–90, Aug. 2019. cellular-connected UAV: Challenges and promising solutions,” arXiv [52] D. Gunduz, P. de Kerret, N. D. Sidiropoulos, D. Gesbert, C. R. Murthy, preprint arXiv:2010.06238, 2020. and M. van der Schaar, “Machine learning in the air,” IEEE J. Sel. [78] P. Chandhar, D. Danev, and E. G. Larsson, “Massive MIMO for Areas Commun., vol. 37, no. 10, pp. 2184–2199, Oct. 2019. communications with drone swarms,” IEEE Trans. Wireless Commun., [53] S. D. Muruganathan, X. Lin, H.-L. Maattanen, J. Sedin, Z. Zou, vol. 17, no. 3, pp. 1604–1629, Mar. 2017. W. A. Hapsari, and S. Yasukawa, “An overview of 3GPP release-15 [79] P. Chandhar and E. G. Larsson, “Massive MIMO for connectivity with study on enhanced LTE support for connected drones,” arXiv preprint drones: Case studies and future directions,” IEEE Access, vol. 7, pp. arXiv:1805.00826, 2018. 94 676–94 691, 2019. 28

[80] A. Garcia-Rodriguez, G. Geraci, D. Lopez-P´ erez,´ L. G. Giordano, for 6G wireless networks: Opportunities, challenges, and trends,” IEEE M. Ding, and E. Bjornson, “The essential guide to realizing 5G- Wireless Commun., 2020. connected UAVs with massive MIMO,” IEEE Commun. Mag., vol. 57, [104] Z. Zhang, L. Lv, Q. Wu, H. Deng, and J. Chen, “Robust and secure no. 12, pp. 84–90, Dec. 2019. communications in intelligent reflecting surface assisted NOMA net- [81] G. Geraci, A. Garcia-Rodriguez, L. G. Giordano, D. Lopez-P´ erez,´ works,” arXiv preprint arXiv:2009.00267, 2020. and E. Bjornson,¨ “Understanding UAV cellular communications: From [105] L. Lv, Q. Wu, Z. Li, N. Al-Dhahir, and J. Chen, “Secure two- existing networks to massive MIMO,” IEEE Access, vol. 6, pp. 67 853– way communications via intelligent reflecting surfaces,” arXiv preprint 67 865, 2018. arXiv:2008.11352, 2020. [82] Y. Huang, Q. Wu, T. Wang, G. Zhou, and R. Zhang, “3D beam tracking [106] Y. Liu, J. Zhao, M. Li, and Q. Wu, “Intelligent reflecting surface for cellular-connected UAV,” IEEE Wireless Commun. Lett., vol. 9, aided MISO uplink communication network: Feasibility and SINR no. 5, pp. 736–740, May 2020. optimization,” arXiv preprint arXiv:2007.01482, 2020. [83] R. Lu, Q. Wu, and R. Zhang, “Pilot decontamination for massive [107] H. Yang, Z. Xiong, J. Zhao, D. Niyato, Q. Wu, H. V. Poor, and MIMO network with UAVs,” arXiv preprint arXiv:2006.05184, 2020. M. Tornatore, “Intelligent reflecting surface assisted anti-jamming com- [84] H. Q. Ngo, A. Ashikhmin, H. Yang, E. G. Larsson, and T. L. Marzetta, munications: A fast reinforcement learning approach,” arXiv preprint “Cell-free massive MIMO versus small cells,” IEEE Trans. Wireless arXiv:2004.12539, 2020. Commun., vol. 16, no. 3, pp. 1834–1850, Mar. 2017. [108] X. Guan, Q. Wu, and R. Zhang, “Joint power control and passive [85] C. DAndrea, A. Garcia-Rodriguez, G. Geraci, L. G. Giordano, and beamforming in IRS-assisted spectrum sharing,” IEEE Commun. Lett., S. Buzzi, “Analysis of UAV communications in cell-free massive vol. 24, no. 7, pp. 1553–1557, Jul. 2020. MIMO systems,” IEEE Open Journal of the Communications Society, [109] “With 1 announcement, the FAA just created an $82 billion market vol. 1, pp. 133–147, 2020. and 100,000 new jobs,” https://www.inc.com/yoram-solomon/with- [86] S. Datta, E. Sharma, and R. Budhiraja, “Power scaling for massive one-rule-the-faa-just-created-an-82-billion-market-and-100000-new- MIMO UAV communication system,” in Proc. IEEE COMSNETS, jobs.html, accessed: Aug. 15, 2018. 2020, pp. 507–510. [110] N. Rajatheva, I. Atzeni, E. Bjornson, A. Bourdoux, S. Buzzi, J.-B. [87] C. D’Andrea, A. Garcia-Rodriguez, G. Geraci, L. G. Giordano, and Dore, S. Erkucuk, M. Fuentes, K. Guan, Y. Hu et al., “White paper S. Buzzi, “Cell-free massive MIMO for UAV communications,” in on broadband connectivity in 6G,” arXiv preprint arXiv:2004.14247, Proc. IEEE ICC Workshops, 2019, pp. 1–6. 2020. [88] E. Bjornson¨ and L. Sanguinetti, “Scalable cell-free massive MIMO [111] N. Rajatheva, I. Atzeni, S. Bicais, E. Bjornson, A. Bourdoux, S. Buzzi, systems,” IEEE Trans. Commun., vol. 68, no. 7, pp. 4247–4261, Jul. C. D’Andrea, J.-B. Dore, S. Erkucuk, M. Fuentes et al., “Scoring 2020. the terabit/s goal: Broadband connectivity in 6G,” arXiv preprint [89] C. Zhang, W. Zhang, W. Wang, L. Yang, and W. Zhang, “Research chal- arXiv:2008.07220, 2020. lenges and opportunities of UAV millimeter-wave communications,” [112] Q. Wu and R. Zhang, “Weighted sum power maximization for intelli- IEEE Wireless Commun., vol. 26, no. 1, pp. 58–62, Feb. 2019. gent reflecting surface aided SWIPT,” IEEE Wireless Commun. Lett., [90] L. Zhang, H. Zhao, S. Hou, Z. Zhao, H. Xu, X. Wu, Q. Wu, and vol. 9, no. 5, pp. 586–590, May 2019. R. Zhang, “A survey on 5G millimeter wave communications for UAV- [113] Q. Wu, X. Zhou, and R. Schober, “IRS-assisted wireless powered assisted wireless networks,” IEEE Access, vol. 7, pp. 117 460–117 504, NOMA: Do we really need different phase shifts in DL and UL?” 2019. IEEE Wireless Commun. Lett., 2021, early access. [91] J. Zhao, F. Gao, L. Kuang, Q. Wu, and W. Jia, “Channel tracking with [114] C. Chaccour, M. N. Soorki, W. Saad, M. Bennis, and P. Popovski, flight control system for UAV mmwave MIMO communications,” IEEE “Risk-based optimization of virtual reality over terahertz reconfigurable Commun. Lett., vol. 22, no. 6, pp. 1224–1227, Jun. 2018. intelligent surfaces,” arXiv preprint arXiv:2002.09052, 2020. [92] P. Zhou, X. Fang, Y. Fang, R. He, Y. Long, and G. Huang, “Beam [115] M. Hua, Q. Wu, D. W. K. Ng, J. Zhao, and L. Yang, “Intelligent management and self-healing for mmwave UAV mesh networks,” IEEE reflecting surface-aided joint processing coordinated multipoint trans- Trans. Veh. Technol., vol. 68, no. 2, pp. 1718–1732, Feb. 2018. mission,” arXiv preprint arXiv:2003.13909, 2020. [93] W. Yuan, C. Liu, F. Liu, S. Li, and D. W. K. Ng, “Learning-based [116] X. Guan, Q. Wu, and R. Zhang, “Intelligent reflecting surface assisted predictive beamforming for UAV communications with jittering,” IEEE secrecy communication: Is artificial noise helpful or not?” IEEE Wireless Commun. Lett., 2020. Wireless Commun. Lett., vol. 9, no. 6, pp. 778–782, 2020. [94] Z. Xiao, P. Xia, and X.-G. Xia, “Enabling UAV cellular with millimeter- [117] V. Jamali, A. M. Tulino, G. Fischer, R. Muller,¨ and R. Schober, wave communication: Potentials and approaches,” IEEE Commun. “Intelligent reflecting and transmitting surface aided millimeter wave Mag., vol. 54, no. 5, pp. 66–73, May 2016. massive MIMO,” arXiv preprint arXiv:1902.07670, 2019. [95] Z. Feng, L. Ji, Q. Zhang, and W. Li, “Spectrum management for [118] B. Zheng, Q. Wu, and R. Zhang, “Intelligent reflecting surface-assisted mmwave enabled UAV swarm networks: Challenges and opportuni- multiple access with user pairing: NOMA or OMA?” IEEE Commun. ties,” IEEE Commun. Mag., vol. 57, no. 1, pp. 146–153, Jan. 2018. Lett., 2020. [96] Q. Wu and R. Zhang, “Towards smart and reconfigurable environment: [119] J. Chen, Y.-C. Liang, Y. Pei, and H. Guo, “Intelligent reflecting surface: Intelligent reflecting surface aided wireless network,” IEEE Commun. A programmable wireless environment for physical layer security,” Mag., vol. 58, no. 1, pp. 106–112, Jan. 2020. IEEE Access, vol. 7, pp. 82 599–82 612, Jul. 2019. [97] S. Gong, X. Lu, D. T. Hoang, D. Niyato, L. Shu, D. I. Kim, and Y.-C. [120] Q. Wu and R. Zhang, “Joint active and passive beamforming opti- Liang, “Towards smart radio environment for wireless communications mization for intelligent reflecting surface assisted SWIPT under QoS via intelligent reflecting surfaces: A comprehensive survey,” arXiv constraints,” IEEE J. Sel. Areas Commun., vol. 38, no. 8, pp. 1735– preprint arXiv:1912.07794, 2019. 1748, Aug. 2020. [98] Q. Wu, S. Zhang, B. Zheng, C. You, and R. Zhang, “Intelligent [121] X. Lu, W. Yang, X. Guan, Q. Wu, and Y. Cai, “Robust and secure reflecting surface aided wireless communications: A tutorial,” arXiv beamforming for intelligent reflecting surface aided mmWave MISO preprint arXiv:2007.02759, 2020. systems,” arXiv preprint arXiv:2003.11195, 2020. [99] E. Basar, “Reconfigurable intelligent surface-based index modulation: [122] H. Lu, Y. Zeng, S. Jin, and R. Zhang, “Aerial intelligent reflecting A new beyond MIMO paradigm for 6G,” IEEE Trans. Commun., surface: Joint placement and passive beamforming design with 3D vol. 68, no. 5, pp. 3187–3196, May 2020. beam flattening,” arXiv preprint arXiv:2007.13295, 2020. [100] Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless [123] Y. Xu, L. Xiao, D. Yang, Q. Wu, and L. Cuthbert, “Throughput network via joint active and passive beamforming,” IEEE Trans. maximization in multi-UAV enabled communication systems with Wireless Commun., vol. 18, no. 11, pp. 5394–5409, Nov. 2019. difference consideration,” IEEE Access, vol. 6, pp. 55 291–55 301, [101] M. Di Renzo, A. Zappone, M. Debbah, M.-S. Alouini, C. Yuen, 2018. J. de Rosny, and S. Tretyakov, “Smart radio environments empowered [124] M. Hua, Y. Wang, Q. Wu, H. Dai, Y. Huang, and L. Yang, “Energy- by reconfigurable intelligent surfaces: How it works, state of research, efficient cooperative secure transmission in multi-UAV-enabled wire- and road ahead,” arXiv preprint arXiv:2004.09352, 2020. less networks,” IEEE Trans. Veh. Technol., vol. 68, no. 8, pp. 7761– [102] Q. Wu and R. Zhang, “Beamforming optimization for wireless network 7775, Aug. 2019. aided by intelligent reflecting surface with discrete phase shifts,” IEEE [125] S. Li, B. Duo, X. Yuan, Y.-C. Liang, and M. Di Renzo, “Reconfigurable Trans. on Commun., vol. 68, no. 3, pp. 1838–1851, Mar. 2020. intelligent surface assisted UAV communication: Joint trajectory design [103] C. Huang, S. Hu, G. C. Alexandropoulos, A. Zappone, C. Yuen, and passive beamforming,” IEEE Wireless Commun. Lett., vol. 9, no. 5, R. Zhang, M. Di Renzo, and M. Debbah, “Holographic MIMO surfaces pp. 716–720, May 2020. 29

[126] Y. Cai, Z. Wei, S. Hu, D. W. K. Ng, and J. Yuan, “Resource allocation [149] J. Zhang, E. Bjornson,¨ M. Matthaiou, D. W. K. Ng, H. Yang, and for power-efficient IRS-assisted UAV communications,” in Proc. IEEE D. J. Love, “Prospective multiple antenna technologies for beyond 5G,” ICC, 2020, pp. 1–7. IEEE J. Sel. Areas Commun., vol. 38, no. 8, pp. 1637–1660, Aug. 2020. [127] L. Wang, K. Wang, C. Pan, W. Xu, and N. Aslam, “Joint trajectory [150] D. W. K. Ng, E. S. Lo, and R. Schober, “Energy-efficient resource and passive beamforming design for intelligent reflecting surface-aided allocation in OFDMA systems with large numbers of base station UAV communications: A deep reinforcement learning approach,” arXiv antennas,” IEEE Trans. Wireless Commun., vol. 11, no. 9, pp. 3292– preprint arXiv:2007.08380, 2020. 3304, Sep. 2012. [128] L. Ge, P. Dong, H. Zhang, J.-B. Wang, and X. You, “Joint beamforming [151] T. L. Marzetta, “Noncooperative cellular wireless with unlimited num- and trajectory optimization for intelligent reflecting surfaces-assisted bers of base station antennas,” IEEE Trans. Wireless Commun., vol. 9, UAV communications,” IEEE Access, vol. 8, pp. 78 702–78 712, 2020. no. 11, pp. 3590–3600, Nov. 2010. [129] D. Ma, M. Ding, and M. Hassan, “Enhancing cellular communications [152] X. Chen, X. Wang, and X. Chen, “Energy-efficient optimization for for UAVs via intelligent reflective surface,” in Proc. IEEE WCNC, wireless information and power transfer in large-scale MIMO systems 2020, pp. 1–6. employing energy beamforming,” IEEE Wireless Commun. Lett., vol. 2, [130] L. Yang, F. Meng, J. Zhang, M. O. Hasna, and M. Di Renzo, “On the no. 6, pp. 667–670, Jun. 2013. performance of RIS-assisted dual-hop UAV communication systems,” [153] L. Zhao, D. W. K. Ng, and J. Yuan, “Multi-user precoding and channel IEEE Trans. Veh. Technol., Early Access. estimation for hybrid millimeter wave systems,” IEEE J. Sel. Areas [131] M. Hua, L. Yang, Q. Wu, C. Pan, C. Li, and A. L. Swindlehurst, Commun., vol. 35, no. 7, pp. 1576–1590, Jul. 2017. “UAV-assisted intelligent reflecting surface symbiotic radio system,” [154] Z. Wei, L. Zhao, J. Guo, D. W. K. Ng, and J. Yuan, “Multi-beam arXiv preprint arXiv:2007.14029, 2020. NOMA for hybrid mmWave systems,” IEEE Trans. Commun., vol. 67, [132] Q. Zhang, W. Saad, and M. Bennis, “Reflections in the sky: Millimeter no. 2, pp. 1705–1719, Feb. 2018. wave communication with UAV-carried intelligent reflectors,” in Proc. IEEE GLOBECOM, 2019, pp. 1–6. [155] Z. Wei, D. W. K. Ng, and J. Yuan, “NOMA for hybrid mmwave IEEE J. Sel. Topics [133] H. Long, M. Chen, Z. Yang, B. Wang, Z. Li, X. Yun, and M. Shikh- communication systems with beamwidth control,” Signal Process. Bahaei, “Reflections in the sky: Joint trajectory and passive beamform- , vol. 13, no. 3, pp. 567–583, Jun. 2019. ing design for secure UAV networks with reconfigurable intelligent [156] D. Nguyen, L.-N. Tran, P. Pirinen, and M. Latva-aho, “Precoding for surface,” arXiv preprint arXiv:2005.10559, 2020. full duplex multiuser MIMO systems: Spectral and energy efficiency [134] T. Shafique, H. Tabassum, and E. Hossain, “Optimization of wireless maximization,” IEEE Trans. Signal Process., vol. 61, no. 16, pp. 4038– relaying with flexible UAV-borne reflecting surfaces,” arXiv preprint 4050, Aug. 2013. arXiv:2006.10969, 2020. [157] Y. Sun, D. W. K. Ng, J. Zhu, and R. Schober, “Multi-objective [135] H. Lu, Y. Zeng, S. Jin, and R. Zhang, “Enabling panoramic full- optimization for robust power efficient and secure full-duplex wireless angle reflection via aerial intelligent reflecting surface,” arXiv preprint communication systems,” IEEE Trans. Wireless Commun., vol. 15, arXiv:2001.07339, 2020. no. 8, pp. 5511–5526, Aug. 2016. [136] S. Zhang and R. Zhang, “Intelligent reflecting surface aided multiple [158] B. Duo, Q. Wu, X. Yuan, and R. Zhang, “Energy efficiency maximiza- access: Capacity region and deployment strategy,” arXiv preprint tion for full-duplex UAV secrecy communication,” IEEE Trans. Veh. arXiv:2002.07091, 2020. Technol., vol. 69, no. 4, pp. 4590–4595, Apr. 2020. [137] S. Jiao, F. Fang, X. Zhou, and H. Zhang, “Joint beamforming and phase [159] C. She, C. Liu, T. Q. S. Quek, C. Yang, and Y. Li, “UAV-assisted shift design in downlink UAV networks with IRS-sssisted NOMA,” uplink transmission for ultra-reliable and low-latency communications,” Journal of Communications and Information Networks, vol. 5, no. 2, in Proc. IEEE ICC Workshops, 2018, pp. 1–6. pp. 138–149, Feb. 2020. [160] A. A. Khuwaja, Y. Chen, N. Zhao, M.-S. Alouini, and P. Dobbins, “A [138] D. Xu, Y. Sun, D. W. K. Ng, and R. Schober, “Multiuser MISO UAV survey of channel modeling for UAV communications,” IEEE Commun. communications in uncertain environments with no-fly zones: Robust Surveys Tuts., vol. 20, no. 4, pp. 2804–2821, Fourth quarter 2018. trajectory and resource allocation design,” IEEE Trans. Commun., [161] R. J. Kerczewski and J. H. Griner, “Control and non-payload commu- vol. 68, no. 5, pp. 3153–3172, May 2020. nications links for integrated unmanned aircraft operations,” NASA, [139] M.-M. Zhao, Q. Wu, M.-J. Zhao, and R. Zhang, “Intelligent reflect- Tech. Rep., Sep. 2012. ing surface enhanced wireless network: Two-timescale beamforming [162] P. Chandhar, D. Danev, and E. G. Larsson, “Massive mimo for optimization,” IEEE Trans. Wireless Commun., 2020. communications with drone swarms,” IEEE Trans. Wireless Commun., [140] S. Abeywickrama, R. Zhang, Q. Wu, and C. Yuen, “Intelligent reflect- vol. 17, no. 3, pp. 1604–1629, Mar. 2017. ing surface: Practical phase shift model and beamforming optimiza- [163] H. Ji, S. Park, J. Yeo, Y. Kim, J. Lee, and B. Shim, “Ultra-reliable and tion,” arXiv preprint arXiv:2002.10112, 2020. low-latency communications in 5G downlink: Physical layer aspects,” [141] Q. Wu and R. Zhang, “Beamforming optimization for intelligent IEEE Wireless Commun., vol. 25, no. 3, pp. 124–130, Mar. 2018. Proc. IEEE ICASSP reflecting surface with discrete phase shifts,” in , [164] P. Popovski, C. Stefanovic, J. J. Nielsen, E. de Carvalho, M. Angjelichi- 2019, pp. 7830–7833. noski, K. F. Trillingsgaard, and A. Bana, “Wireless access in ultra- [142] M.-M. Zhao, Q. Wu, M.-J. Zhao, and R. Zhang, “Exploiting amplitude reliable low-latency communication (URLLC),” IEEE Trans. Commun., control in intelligent reflecting surface aided wireless communication vol. 67, no. 8, pp. 5783–5801, Aug. 2019. with imperfect CSI,” arXiv preprint arXiv:2005.07002, 2020. [165] A. Anand, G. de Veciana, and S. Shakkottai, “Joint scheduling of [143] Q. Zhang, Y.-C. Liang, and H. V. Poor, “Large intelligent sur- URLLC and eMBB traffic in 5G wireless networks,” IEEE/ACM Trans. face/antennas (LISA) assisted symbiotic radio for IoT communica- Netw., vol. 28, no. 2, pp. 477–490, Feb. 2020. tions,” arXiv preprint arXiv:2002.00340, 2020. [144] W. Yan, X. Yuan, and X. Kuai, “Passive beamforming and information [166] Y. Sun, D. Xu, D. W. K. Ng, L. Dai, and R. Schober, “Optimal transfer via large intelligent surface,” IEEE Wireless Commun. Lett., 3D-trajectory design and resource allocation for solar-powered UAV vol. 9, no. 4, pp. 533–537, Apr. 2020. communication systems,” IEEE Trans. Commun., vol. 67, no. 6, pp. [145] G. J. Sutton, J. Zeng, R. P. Liu, W. Ni, D. N. Nguyen, B. A. 4281–4298, Jun. 2019. Jayawickrama, X. Huang, M. Abolhasan, Z. Zhang, E. Dutkiewicz, [167] R. Li, Z. Wei, L. Yang, D. W. K. Ng, J. Yuan, and J. An, “Resource and T. Lv, “Enabling technologies for ultra-reliable and low latency allocation for secure multi-UAV communication systems with multi- communications: From PHY and MAC layer perspectives,” IEEE eavesdropper,” IEEE Trans. Commun., vol. 68, no. 7, pp. 4490–4506, Commun. Surveys Tuts., vol. 21, no. 3, pp. 2488–2524, Third quarter Jul. 2020. 2019. [168] Y. Cai, Z. Wei, R. Li, D. W. K. Ng, and J. Yuan, “Joint trajectory and [146] K. S. Kim, D. K. Kim, C. Chae, S. Choi, Y. Ko, J. Kim, Y. Lim, resource allocation design for energy-efficient secure UAV communi- M. Yang, S. Kim, B. Lim, K. Lee, and K. L. Ryu, “Ultrareliable and cation systems,” IEEE Trans. Commun., vol. 68, no. 7, pp. 4536–4553, low-latency communication techniques for tactile internet services,” Jul. 2020. Proc. IEEE, vol. 107, no. 2, pp. 376–393, Feb. 2019. [169] X. Pang, Z. Li, X. Chen, Y. Cao, N. Zhao, Y. Chen, and Z. Ding, “UAV- [147] “Technical Specification Group Services and System Aspects; Service aided NOMA networks with optimization of trajectory and precoding,” requirements for the 5G system; Stage 1 (Release 16),” 3GPP TS in Proc. IEEE WCSP, 2018, pp. 1–6. 22.261 V16.12.0 (2020-07), Tech. Rep., Jul 2020. [170] C. She, C. Liu, T. Q. S. Quek, C. Yang, and Y. Li, “Ultra-reliable and [148] V. W. S. Wong, R. Schober, D. W. K. Ng, and L.-C. Wang, Key low-latency communications in unmanned aerial vehicle communica- Technologies for 5G Wireless Systems. Cambridge University Press, tion systems,” IEEE Trans. Commun., vol. 67, no. 5, pp. 3768–3781, 2017. Oct. 2019. 30

[171] C. Pan, H. Ren, Y. Deng, M. Elkashlan, and A. Nallanathan, “Joint networks: Research challenges and future trends,” IEEE J. Sel. Areas blocklength and location optimization for URLLC-enabled UAV relay Commun., vol. 35, no. 10, pp. 2181–2195, Oct. 2017. systems,” IEEE Commun. Lett., vol. 23, no. 3, pp. 498–501, Mar. 2019. [193] W. Mei and R. Zhang, “Uplink cooperative NOMA for cellular- [172] H. Ren, C. Pan, K. Wang, W. Xu, M. Elkashlan, and A. Nallanathan, connected UAV,” IEEE J. Sel. Topics Signal Process., vol. 13, no. 3, “Joint transmit power and placement optimization for urllc-enabled pp. 644–656, Jun. 2019. UAV relay systems,” IEEE Trans. Veh. Technol., vol. 69, no. 7, pp. [194] E. Dahlman, S. Parkvall, and J. Skold, 4G: LTE/LTE-advanced for 8003–8007, Jul, 2020. mobile broadband. Academic press, 2013. [173] W. R. Ghanem, V. Jamali, Y. Sun, and R. Schober, “Resource allocation [195] W. Mei and R. Zhang, “Uplink cooperative interference cancellation for multi-user downlink URLLC-OFDMA systems,” Proc. IEEE ICC for cellular-connected UAV: A quantize-and-forward approach,” IEEE Workshops, pp. 1–6, 2019. Wireless Commun. Lett., 2020. [174] Y. Polyanskiy, H. V. Poor, and S. Verdu,´ “Channel coding rate in the [196] ——, “Cooperative downlink interference transmission and cancel- finite blocklength regime,” IEEE Trans. Inf. Theory, vol. 56, no. 5, pp. lation for cellular-connected UAV: A divide-and-conquer approach,” 2307–2359, May 2010. IEEE Trans. Commun., vol. 68, no. 2, pp. 1297–1311, Feb. 2019. [175] M. Shirvanimoghaddam, M. S. Mohammadi, R. Abbas, A. Minja, [197] W. K. New, C. Y. Leow, K. Navaie, and Z. Ding, “Robust non- C. Yue, B. Matuz, G. Han, Z. Lin, W. Liu, Y. Li, S. Johnson, and orthogonal multiple access for aerial and ground users,” IEEE Trans. B. Vucetic, “Short block-length codes for ultra-reliable low latency Wireless Commun., vol. 19, no. 7, pp. 4793–4805, Jul. 2020. communications,” IEEE Commun. Mag., vol. 57, no. 2, pp. 130–137, [198] Q. Wu, J. Xu, and R. Zhang, “Capacity characterization of UAV- Feb. 2019. enabled two-user broadcast channel,” IEEE J. Sel. Areas Commun., [176] X. Sun, C. Shen, D. W. K. Ng, and Z. Zhong, “Robust trajectory and vol. 36, no. 9, pp. 1955–1971, Sep. 2018. resource allocation design for secure UAV-aided communications,” in Proc. IEEE ICC Workshops, 2019, pp. 1–6. [199] M. F. Sohail, C. Y. Leow, and S. Won, “Non-orthogonal multiple access [177] R. Fan, J. Cui, S. Jin, K. Yang, and J. An, “Optimal node placement for unmanned aerial vehicle assisted communication,” IEEE Access, and resource allocation for UAV relaying network,” IEEE Commun. vol. 6, pp. 22 716–22 727, 2018. Lett., vol. 22, no. 4, pp. 808–811, Apr. 2018. [200] P. K. Sharma and D. I. Kim, “UAV-enabled downlink wireless system [178] W. R. Ghanem, V. Jamali, Y. Sun, and R. Schober, “Resource alloca- with non-orthogonal multiple access,” in Proc. IEEE Globecom, 2017, tion for multi-user downlink MISO OFDMA-URLLC systems,” IEEE pp. 1–6. Trans. Commun., Early Access. [201] A. A. Nasir, H. D. Tuan, T. Q. Duong, and H. V. Poor, “UAV-enabled [179] M. Shirvanimoghaddam, M. S. Mohammadi, R. Abbas, A. Minja, communication using NOMA,” IEEE Trans. Commun., vol. 67, no. 7, C. Yue, B. Matuz, G. Han, Z. Lin, W. Liu, Y. Li, S. Johnson, and pp. 5126–5138, Jul. 2019. B. Vucetic, “Short block-length codes for ultra-reliable low latency [202] T. Hou, Y. Liu, Z. Song, X. Sun, and Y. Chen, “Multiple antenna communications,” IEEE Commun. Mag., vol. 57, no. 2, pp. 130–137, aided NOMA in UAV networks: A stochastic geometry approach,” 2019. IEEE Trans. Commun., vol. 67, no. 2, pp. 1031–1044, Feb. 2018. [180] X. Chen, D. W. K. Ng, W. Yu, E. G. Larsson, N. Al-Dhahir, and [203] ——, “Exploiting NOMA for UAV communications in large-scale R. Schober, “Massive access for 5G and beyond,” arXiv preprint cellular networks,” IEEE Trans. Commun., vol. 67, no. 10, pp. 6897– arXiv:2002.03491, 2020. 6911, Oct. 2019. [181] X. Chen, Massive Access for Cellular Internet of Things Theory and [204] X. Liu, J. Wang, N. Zhao, Y. Chen, S. Zhang, Z. Ding, and F. R. Yu, Technique. Springer, 2019. “Placement and power allocation for NOMA-UAV networks,” IEEE [182] J. Zhu, D. W. K. Ng, N. Wang, R. Schober, and V. K. Bhargava, Wireless Commun. Lett.,, vol. 8, no. 3, pp. 965–968, Jun. 2019. “Analysis and design of secure massive MIMO systems in the presence [205] R. Duan, J. Wang, C. Jiang, H. Yao, Y. Ren, and Y. Qian, “Resource of hardware impairments,” IEEE Trans. Wireless Commun., vol. 16, allocation for multi-UAV aided IoT NOMA uplink transmission sys- no. 3, pp. 2001–2016, Mar. 2017. tems,” IEEE Internet Things J., vol. 6, no. 4, pp. 7025–7037, Aug. [183] “Cisco visual networking index: Global mobile data traffic forecast,” 2019. Cisco, Tech. Rep., Feb 2017. [206] N. Rupasinghe, Y. Yapıcı, I. Guvenc¸,¨ M. Ghosh, and Y. Kakishima, [184] A. Orsino, A. Ometov, G. Fodor, D. Moltchanov, L. Militano, S. An- “Angle feedback for NOMA transmission in mmwave drone networks,” dreev, O. N. C. Yilmaz, T. Tirronen, J. Torsner, G. Araniti, A. Iera, IEEE J. Sel. Topics Signal Process., vol. 13, no. 3, pp. 628–643, Jun. M. Dohler, and Y. Koucheryavy, “Effects of heterogeneous mobility on 2019. D2D- and drone-assisted mission-critical MTC in 5G,” IEEE Commun. [207] W. Bin, L. Wenxin, and L. Liu, “A survey of energy conservation, Mag, vol. 55, no. 2, pp. 79–87, Feb. 2017. routing and coverage in wireless sensor networks,” in International [185] L. Liu, E. G. Larsson, W. Yu, P. Popovski, C. Stefanovic, and E. de Conference on Active Media Technology. Springer, 2011, pp. 59–70. Carvalho, “Sparse signal processing for grant-free massive connectiv- [208] Q. Wu, M. Tao, and W. Chen, “Joint Tx/Rx energy-efficient scheduling ity: A future paradigm for random access protocols in the internet of in multi-radio wireless networks: A divide-and-conquer approach,” things,” IEEE Signal Process. Mag., vol. 35, no. 5, pp. 88–99, May IEEE Trans. Wireless Commun., vol. 15, no. 4, pp. 2727–2740, Apr. 2018. 2015. [186] Z. Wei, L. Yang, D. W. K. Ng, J. Yuan, and L. Hanzo, “On the [209] F. K. Shaikh and S. Zeadally, “Energy harvesting in wireless sensor performance gain of NOMA over OMA in uplink communication networks: A comprehensive review,” Renewable and Sustainable En- systems,” IEEE Trans. Commun., vol. 68, no. 1, pp. 536–568, Jan. ergy Reviews, vol. 55, pp. 1041–1054, 2016. 2020. [187] X. Chen, Z. Zhang, C. Zhong, R. Jia, and D. W. K. Ng, “Fully non- [210] G. Zhang, J. Xu, Q. Wu, M. Cui, X. Li, and F. Lin, “Wireless powered IEEE Trans. Veh. orthogonal communication for massive access,” IEEE Trans. Commun., cooperative jamming for secure OFDM system,” Technol. vol. 66, no. 4, pp. 1717–1731, Apr. 2017. , vol. 67, no. 2, pp. 1331–1346, Feb. 2018. [188] Q. Wu, M. Tao, D. W. K. Ng, W. Chen, and R. Schober, “Energy- [211] Q. Wu, W. Chen, D. W. K. Ng, and R. Schober, “Spectral and energy efficient resource allocation for wireless powered communication net- efficient wireless powered IoT networks: NOMA or TDMA?” IEEE works,” IEEE Tran. Wireless Commun., vol. 15, no. 3, pp. 2312–2327, Trans. Veh. Technol., vol. 67, no. 7, pp. 6663–6667, Jul. 2018. Mar. 2015. [212] Q. Wu, W. Chen, and J. Li, “Wireless powered communications with [189] X. Chen, X. Wang, and X. Chen, “Energy-efficient optimization for initial energy: QoS guaranteed energy-efficient resource allocation,” wireless information and power transfer in large-scale MIMO systems IEEE Commun. Lett., vol. 19, no. 12, pp. 2278–2281, Dec. 2015. employing energy beamforming,” IEEE Wireless Commun. Lett., vol. 2, [213] Q. Wu, G. Zhang, D. W. K. Ng, W. Chen, and R. Schober, “Generalized no. 6, pp. 667–670, Jun. 2013. wireless-powered communications: When to activate wireless power [190] Z. Hasan, G. Bansal, E. Hossain, and V. Bhargava, “Energy-efficient transfer?” IEEE Trans. Veh. Technol., vol. 68, no. 8, pp. 8243–8248, power allocation in OFDM-based cognitive radio systems: A risk-return Aug. 2019. model,” IEEE Trans. Wireless Commun., vol. 8, no. 12, pp. 6078–6088, [214] Q. Wu, W. Chen, D. W. K. Ng, J. Li, and R. Schober, “User-centric Dec. 2009. energy efficiency maximization for wireless powered communications,” [191] C. Zhan, Y. Zeng, and R. Zhang, “Energy-efficient data collection in IEEE Trans. Wireless Commun., vol. 15, no. 19, pp. 6898–6912, Jul. UAV enabled wireless sensor network,” IEEE Wireless Commun. Lett., 2016. vol. 7, no. 3, pp. 328–331, Jun. 2018. [215] J. Xu, Y. Zeng, and R. Zhang, “UAV-enabled wireless power transfer: [192] Z. Ding, X. Lei, G. K. Karagiannidis, R. Schober, J. Yuan, and trajectory design and energy region characterization,” in 2017 IEEE V. K. Bhargava, “A survey on non-orthogonal multiple access for 5G Globecom Workshops (GC Wkshps), 2018. 31

[216] J. Xu, Y. Zeng, and R. Zhang, “UAV-enabled wireless power transfer: [239] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Mobile internet trajectory design and energy optimization,” IEEE Trans. Wireless of things: Can UAVs provide an energy-efficient mobile architecture?” Commun., vol. 17, no. 8, pp. 5092–5106, Aug. 2018. in Proc. IEEE GLOBECOM, 2016, pp. 1–6. [217] Y. Hu, X. Yuan, J. Xu, and A. Schmeink, “Optimal 1D trajectory [240] Z. Kaleem, M. Yousaf, A. Qamar, A. Ahmad, T. Q. Duong, W. Choi, design for UAV-enabled multiuser wireless power transfer,” IEEE and A. Jamalipour, “UAV-empowered disaster-resilient edge architec- Trans. Commun., vol. 67, no. 8, pp. 5674–5688, Aug. 2019. ture for delay-sensitive communication,” IEEE Netw., vol. 33, no. 6, [218] Z. Yang, W. Xu, and M. Shikh-Bahaei, “Energy efficient UAV commu- pp. 124–132, 2019. nication with energy harvesting,” IEEE Trans. Veh. Technol., vol. 69, [241] M. Hua, Y. Wang, Z. Zhang, C. Li, Y. Huang, and L. Yang, “Power- no. 2, pp. 1913–1927, Feb. 2020. efficient communication in UAV-aided wireless sensor networks,” IEEE [219] L. Xie, J. Xu, and R. Zhang, “Throughput maximization for UAV- Commun. Lett., vol. 22, no. 6, pp. 1264–1267, Jun. 2018. enabled wireless powered communication networks,” IEEE Internet [242] M. Hua, L. Yang, Q. Wu, and A. L. Swindlehurst, “3D UAV trajectory Things J., vol. 6, no. 2, pp. 1690–1703, Apr. 2019. and communication design for simultaneous uplink and downlink [220] L. Xie, J. Xu, and Y. Zeng, “Common throughput maximization for transmission,” IEEE Trans. Commun., vol. 68, no. 9, pp. 5908–5923, UAV-enabled interference channel with wireless powered communi- Sep. 2020. cations,” IEEE Trans. Commun., vol. 68, no. 5, pp. 3197–3212, May [243] Y. Wu, W. Yang, X. Guan, and Q. Wu, “Energy-efficient trajectory 2020. design for UAV-enabled communication under malicious jamming,” [221] J. Tang, J. Song, J. Ou, J. Luo, X. Zhang, and K. Wong, “Minimum arXiv preprint arXiv:2009.06161, 2020. throughput maximization for multi-UAV enabled WPCN: A deep [244] M. Li, N. Cheng, J. Gao, Y. Wang, L. Zhao, and X. Shen, “Energy- reinforcement learning method,” IEEE Access, vol. 8, pp. 9124–9132, efficient UAV-assisted mobile edge computing: Resource allocation and 2020. trajectory optimization,” IEEE Trans. Veh. Technol., vol. 69, no. 3, pp. [222] J. Park, H. Lee, S. Eom, and I. Lee, “UAV-aided wireless powered 3424–3438, Mar. 2020. communication networks: Trajectory optimization and resource alloca- [245] G. Yang, R. Dai, and Y. Liang, “Energy-efficient UAV backscatter tion for minimum throughput maximization,” IEEE Access, vol. 7, pp. communication with joint trajectory and resource optimization,” in 134 978–134 991, 2019. Proc. IEEE ICC, 2019, pp. 1–6. [223] F. Wu, D. Yang, L. Xiao, and L. Cuthbert, “Minimum-throughput [246] J. Kennedy and R. Eberhart, “Particle swarm optimization,” in Proc. maximization for multi-UAV-enabled wireless-powered communica- Proceedings of international conference on neural networks, vol. 4. tion networks,” Sensors, vol. 19, no. 7, p. 1491, 2019. IEEE, 1995, pp. 1942–1948. [224] F. Huang, J. Chen, H. Wang, G. Ding, Z. Xue, Y. Yang, and F. Song, [247] J.-J. Kim and J.-J. Lee, “Trajectory optimization by particle swarm “UAV-assisted SWIPT in internet of things with power splitting: optimization in motion planning,” in Progress in Systems Engineering. Trajectory design and power allocation,” IEEE Access, vol. 7, pp. Springer, 2015, pp. 299–305. 68 260–68 270, 2019. [248] V. Roberge, M. Tarbouchi, and G. Labonte,´ “Comparison of parallel [225] J. Kang and C. Chun, “Joint trajectory design, tx power allocation, genetic algorithm and particle swarm optimization for real-time UAV and rx power splitting for UAV-enabled multicasting SWIPT systems,” path planning,” IEEE Trans. on industrial informatics, vol. 9, no. 1, IEEE Systems J., vol. 14, no. 3, pp. 3740–3743, 2020. pp. 132–141, Feb. 2012. [226] W. Feng, J. Tang, Y. Yu, J. Song, N. Zhao, G. Chen, K.-K. Wong, and [249] J. L. Foo, J. Knutzon, J. Oliver, and E. Winer, “Three-dimensional path J. Chambers, “UAV-enabled SWIPT in IoT networks for emergency planning of unmanned aerial vehicles using particle swarm optimiza- communications,” IEEE Wireless Commun., vol. 27, no. 5, pp. 140– tion,” in Proc. AIAA/ISSMO multidisciplinary analysis and optimization 147, Oct. 2020. conference, 2006, p. 6995. [227] S. Morton, R. D’Sa, and N. Papanikolopoulos, “Solar powered UAV: [250] J. Liu, X. Wang, B. Bai, and H. Dai, “Age-optimal trajectory planning Design and experiments,” in Proc. IEEE/RSJ IROS, 2015, pp. 2460– for uav-assisted data collection,” arXiv preprint arXiv:1804.09356, 2466. 2018. [228] P. Oettershagen, A. Melzer, T. Mantel, K. Rudin, T. Stastny, [251] M. Gemeinder and M. Gerke, “GA-based path planning for mobile B. Wawrzacz, T. Hinzmann, K. Alexis, and R. Siegwart, “Perpetual robot systems employing an active search algorithm,” Applied Soft flight with a small solar-powered UAV: Flight results, performance Computing, vol. 3, no. 2, pp. 149–158, 2003. analysis and model validation,” in Proc. IEEE Aerospace Conference, 2016, pp. 1–8. [252] L. Li, Q. Cheng, K. Xue, C. Yang, and Z. Han, “Downlink transmit [229] S. Hu, Q. Wu, and X. Wang, “Energy management and trajectory power control in ultra-dense UAV network based on mean field game optimization for UAV-enabled legitimate monitoring systems,” IEEE and deep reinforcement learning,” IEEE Trans. Veh. Technol., vol. 69, Trans. Wireless Commun., Early Access. no. 12, pp. 15 594–15 605, Dec. 2020. [230] J. A. Duffie, W. A. Beckman, and W. M. Worek, Solar Engineering of [253] Y. Kang, S. Liu, H. Zhang, W. Li, Z. Han, S. Osher, and H. V. Thermal Processes, 2nd ed. Wiley,, 1991. Poor, “Joint sensing task assignment and collision-free trajectory [231] Y. Sun, D. Xu, D. W. K. Ng, L. Dai, and R. Schober, “Optimal optimization for mobile vehicle networks using mean-field games,” 3D-trajectory design and resource allocation for solar-powered UAV IEEE Internet of Things Journal, pp. 1–1, 2020. communication systems,” IEEE Trans. Commun., vol. 67, no. 6, pp. [254] H. Zhang, Z. Han, and H. V. Poor, “Trajectory optimization for UAV- 4281–4298, Jun. 2019. to-device underlaid cellular networks by mean-field-type control,” in [232] J. Ouyang, Y. Che, J. Xu, and K. Wu, “Throughput maximization for Proc. IEEE GLOBECOM, 2020, pp. 1–6. laser-powered UAV wireless communication systems,” in Proc. IEEE [255] H. Zhang, L. Song, Z. Han, and H. V. Poor, “Cooperation techniques ICC Workshops, 2018, pp. 1–6. for a cellular internet of unmanned aerial vehicles,” IEEE Wireless [233] M. Zhao, Q. Shi, and M. Zhao, “Efficiency maximization for UAV- Commun., vol. 26, no. 5, pp. 167–173, Oct. 2019. enabled mobile relaying systems with laser charging,” IEEE Trans. [256] Z. Ding, R. Schober, P. Fan, and H. V. Poor, “Simple semi-grant-free Wireless Commun., vol. 19, no. 5, pp. 3257–3272, May 2020. transmission strategies assisted by non-orthogonal multiple access,” [234] M. Lahmeri, M. A. Kishk, and M. Alouini, “Stochastic geometry-based IEEE Trans. on Commun., vol. 67, no. 6, pp. 4464–4478, Jun. 2019. analysis of airborne base stations with laser-powered UAVs,” IEEE [257] J. Choi, J. Ding, N. P. Le, and Z. Ding, “Grant-free random access Commun. Lett., vol. 24, no. 1, pp. 173–177, Jan. 2020. in machine-type communication: Approaches and challenges,” arXiv [235] M. Z. Hassan, M. J. Hossain, J. Cheng, and V. C. M. Leung, preprint arXiv:2012.10550, 2020. “Statistical-QoS guarantee for IoT network driven by laser-powered [258] Q. Wu, W. Mei, and R. Zhang, “Safeguarding wireless network with UAV relay and RF backscatter communications,” IEEE Trans. on Green UAVs: a physical layer security perspective,” IEEE Wireless Commun., Commun. and Netw., vol. 5, no. 1, pp. 406–425, Mar. 2021. vol. 26, no. 5, pp. 12–18, Oct. 2019. [236] Y. Zeng and R. Zhang, “Energy-efficient UAV communication with [259] G. Zhang, Q. Wu, M. Cui, and R. Zhang, “Securing UAV communi- trajectory optimization,” IEEE Trans. Wireless Commun., vol. 16, no. 6, cations via joint trajectory and power control,” IEEE Trans. Wireless pp. 3747–3760, Jun. 2017. Commun., vol. 18, no. 2, pp. 1376–1389, Feb. 2019. [237] Y. Zeng, J. Xu, and R. Zhang, “Energy minimization for wireless com- [260] A. Li, Q. Wu, and R. Zhang, “UAV-enabled cooperative jamming munication with rotary-wing UAV,” IEEE Trans. Wireless Commun., for improving secrecy of ground wiretap channel,” IEEE Wireless vol. 18, no. 4, pp. 2329–2345, Apr. 2019. Commun. Lett., vol. 8, no. 1, pp. 181–184, Feb. 2019. [238] M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Optimal transport [261] M. Cui, G. Zhang, Q. Wu, and D. W. K. Ng, “Robust trajectory and theory for power-efficient deployment of unmanned aerial vehicles,” in transmit power design for secure UAV communications,” IEEE Trans. Proc. IEEE ICC, 2016, pp. 1–6. Veh. Technol., vol. 67, no. 9, pp. 9042–9046, Sep. 2018. 32

[262] X. Zhou, Q. Wu, S. Yan, F. Shu, and J. Li, “UAV-enabled secure [287] Y. L. Sit, C. Sturm, and T. Zwick, “Doppler estimationin an OFDM communications: joint trajectory and transmit power optimization,” joint radar and communication systems,” in 2011 German Microwave IEEE Trans. Veh. Technol., vol. 68, no. 4, pp. 4069–4073, Apr. 2019. Conf., pp. 1–4, Mar. 2011. [263] Y. Wu, W. Yang, X. Guan, and Q. Wu, “Energy-efficient trajectory [288] M. Braun, C. Sturm, A. Niethammer, and F. K. Jondral, “Parameter- design for uav-enabled communication under malicious jamming,” ization of joint OFDM-coded radar signals,” in Proc. IEEE PIMRC, IEEE Wireless Commun. Lett., pp. 1–1, 2020. 2009. [264] X. Yu and Y. Zhang, “Sense and avoid technologies with applications [289] G. Franken, H. Nikookar, and P. V. Genderen, “Doppler tolerance of to unmanned aircraft systems: Review and prospects,” Progress in OFDM-coded radar signals,” in Proc. EuRAD, 2006. Aerospace Sciences, vol. 74, pp. 152–166, Jan. 2015. [290] X. Liu, T. Huang, N. Shlezinger, Y. Liu, J. Zhou, and Y. C. Eldar, [265] I. Guvenc, F. Koohifar, S. Singh, M. L. Sichitiu, and D. Matolak, “De- “Joint transmit beamforming for multiuser MIMO communications and tection, tracking, and interdiction for amateur drones,” IEEE Commun. MIMO radar,” IEEE Trans. Signal Process., vol. 68, pp. 3929–3944, Mag., vol. 56, no. 4, pp. 75–81, Apr. 2018. 2020. [266] F. Fioranelli et al, “Classification of loaded/unloaded micro-drones [291] J. A. Zhang, X. Huang, Y. J. Guo, J. Yuan, and R. W. Heath, using multistatic radar,” IEEE Electron. Lett., vol. 51, no. 22, pp. 1813– “Multibeam for joint communication and radar sensing using steerable 1815, Oct. 2015. analog antenna arrays,” IEEE Trans. Veh. Tecnol., vol. 68, no. 1, pp. [267] N. Mohajerin, J. Histon, R. Dizaji, and S. L. Waslande, “Feature 671–685, Jan. 2019. extraction and radar track classification for detecting UAVs in civillian [292] Y. Zeng and X. Xu, “Towards environment-aware 6G communications IEEE Radar Conf. airspace,” , pp. 0674–0679, May 2014. via channel knowledge map,” arXiv preprint arXiv:2007.09332, 2020. [268] P. Hugler, F. Roos, M. Schartel, M. Geiger, and C. Waldschmidt, “Radar [293] Z. Zhou, X. Chen, E. Li, L. Zeng, K. Luo, and J. Zhang, “Edge taking off: New capabilities for UAVs,” IEEE Microw. Mag., vol. 19, intelligence: Paving the last mile of artificial intelligence with edge no. 7, pp. 43–53, Jul. 2018. computing,” Proc. IEEE, vol. 107, no. 8, pp. 1738–1762, 2019. [269] G. Yang, et al, “Unmanned aerial vehicle remote sensing for field- based crop phenotyping: Current status and perspectives,” Front. Plant [294] G. Zhu, D. Liu, Y. Du, C. You, J. Zhang, and K. Huang, “Toward Sci., vol. 8, p. 1111, Jan. 2017. an intelligent edge: Wireless communication meets machine learning,” [270] H. Essen, W. Johannes, S. Stanko, R. Sommer, A. Wahlen, and J. IEEE Commun. Mag., vol. 58, no. 1, pp. 19–25, Jan. 2020. Wilcke, “High resolution W-band UAV SAR,” in Proc. IEEE Int. [295] Y. Mao, C. You, J. Zhang, K. Huang, and K. B. Letaief, “A survey Geoscience and Remote Sensing Symp., pp. 5033–5036, 2012. on mobile edge computing: The communication perspective,” IEEE [271] F. Koohifar, A. Kumbhar, and I. Guvenc, “Receding horizon multi- Commun. Surveys Tuts., vol. 19, no. 4, pp. 2322–2358, Fourth quarter UAV cooperative tracking of moving RF source,” IEEE Commun. Lett., 2017. vol. 21, no. 6, pp. 1433–1436, Jun. 2017. [296] F. Wang, J. Xu, X. Wang, and S. Cui, “Joint offloading and computing [272] A. Guerra, D. Dardari, and P. M. Djuric, “Dynamic radar networks of optimization in wireless powered mobile-edge computing systems,” UAVs: A tutorial overview and tracking performance comparison with IEEE Trans. Wireless Commun., vol. 17, no. 3, pp. 1784–1797, Mar. terrestrial radar networks,” IEEE Veh. Technol. Mag., vol. 15, no. 2, 2018. pp. 113–120, Jun. 2020. [297] V. Kouhdaragh, F. Verde, G. Gelli, and J. Abouei, “On the application [273] Z. Xiao and Y. Zeng, “An overview on integrated localization and of machine learning to the design of UAV-based 5G radio access communication towards 6G,” arXiv preprint arXiv:2006.01535, 2020. networks,” Electronics, vol. 9, no. 4, p. 689, Feb. 2020. [274] Z. Hu, Z. Bai, Y. Yang, Z. Zheng, K. Bian, and L. Song, “UAV aided [298] P. S. Bithas, E. T. Michailidis, N. Nomikos, D. Vouyioukas, and A. G. aerial-ground iot for air quality sensing in smart city: Architecture, Kanatas, “A survey on machine-learning techniques for UAV-based technologies, and implementation,” IEEE Netw., vol. 33, no. 2, pp. communications,” Sensors, vol. 19, no. 23, p. 5170, Nov. 2019. 14–22, Apr. 2019. [299] R. S. Sutton and A. G. Barto, Introduction to Reinforcement Learning, [275] M. I. Skolnik, Radar handbook, 3rd ed. New York: McGraw-Hill, 2nd ed. Cambridge, Massachusetts: MIT Press, 2017. 2008. [300] X. Liu, M. Chen, Y. Liu, Y. Chen, S. Cui, and L. Hanzo, “Artificial [276] S. M. Kay, Fundumentals of Statistical Signal Processing: Estimation intelligence aided next-generation networks relying on UAVs,” arXiv Theory. New Jersey: Prentice-Hall, 1993. preprint arXiv:2001.11958, 2020. [277] A. R. Chiriyath, B. Paul, and D. W. Bliss, “Radar-commuications [301] H. Bayerlein, P. D. Kerret, and D. Gesbert, “Trajectory optimization for convergence: Coexistence, cooperation, and co-design,” IEEE Trans. autonomous flying base station via reinforcement learning,” in Proc. Cognitive Commun. Netw., vol. 3, no. 1, pp. 1–12, Mar. 2017. IEEE SPAWC, Jun. 2018, pp. 1–5. [278] D. W. Bliss and K. W. Forsythe, “Multiple-input multiple-output [302] Y. Huang, X. Mo, J. Xu, L. Qiu, and Y. Zeng, “Online maneuver design (MIMO) radar and imaging: degrees of freedom and resolution,” in for UAV-enabled NOMA systems via reinforcement learning,” in Proc. Asilomar Conf. on Signals, Systems Computers, pp. 54–59, Nov. 2003. IEEE WCNC, 2020, pp. 1–6. [279] J. Li. and P. Stoica, MIMO radar signal processing. Hoboken: Wiley, [303] H. Bayerlein, M. Theile, M. Caccamo, and D. Gesbert, “UAV path 2008. planning for wireless data harvesting: A deep reinforcement learning [280] D. Ma, N. Shlezinger, T. Huang, Y. Liu, and Y. C. Eldar, “Joint radar- approach,” arXiv preprint arXiv:2007.00544, 2020. communications strategies for autonomous vehicles,” arXiv preprint [304] X. Liu, Y. Liu, and Y. Chen, “Reinforcement learning in multiple- arXiv:1909.01729, 2019. UAV networks: Deployment and movement design,” IEEE Trans. Veh. [281] D. Solomitckii, M. Gapeyenko, V. Semkin, S. Andreev, and Y. Kouch- Technol., vol. 68, no. 8, pp. 8036–8049, Aug. 2019. eryavy, “Technologies for efficient amateur drone detection in 5G millimeter-wave cellular infrastructure,” IEEE Commun. Mag., vol. 56, [305] C. H. Liu, X. Ma, X. Gao, and J. Tang, “Distributed energy-efficient no. 1, pp. 43–50, Apr. 2018. multi-UAV navigation for long-term communication coverage by deep IEEE Trans. Mobile Computing [282] L. Zheng, M. Lops, Y. C. Eldar, and X. Wang, “Radar and communi- reinforcement learning,” , vol. 19, no. 6, cation coexistence: An overview: A review of recent methods,” IEEE pp. 1274–1285, Jun. 2020. Signal Process. Mag., vol. 36, no. 5, pp. 85–99, Sep. 2019. [306] M. Chen, W. Saad, and C. Yin, “Liquid state machine learning for [283] A. Hassanien, M. G. Amin, E. Aboutanios, and B. Himed, “Dual- resource and cache management in LTE-U unmanned aerial vehicle function radar communication systems: A solution to the spectrum (UAV) networks,” IEEE Trans. Wireless Commun., vol. 18, no. 3, pp. congestion problem,” IEEE Signal Mag., vol. 36, no. 5, pp. 115–126, 1504–1517, Mar. 2019. Sep. 2019. [307] J. Hu, H. Zhang, L. Song, R. Schober, and H. V. Poor, “Cooperative [284] C. Sahin, J. Jakabosky, P. M. McCormick, J. G. Metcalf, and S. D. internet of UAVs: Distributed trajectory design by multi-agent deep Blunt, “A novel approach for embedding communication symbols into reinforcement learning,” IEEE Trans. Commun., Early Access. physical radar waveforms,” in Proc. IEEE RadarConf, pp. 1498–1503, [308] Y. Zeng, X. Xu, S. Jin, and R. Zhang, “Simultaneous navigation and May 2017. radio mapping for cellular-connected UAV with deep reinforcement [285] A. Hassanien, M. G. Amin, Y. D. Zhang, and F. Ahmad, “Dual-function learning,” arXiv preprint arXiv:2003.07574, 2020. radar-communications: Information embedding using sidelobe control [309] E. Fonseca, B. Galkin, L. A. DaSilva, and I. Dusparic, “Adaptive and waveform diversity,” IEEE Trans. Signal Process., vol. 64, no. 8, height optimisation for cellular-connected UAVs using reinforcement pp. 2168–2181, Apr. 2016. learning,” arXiv preprint arXiv:2007.13695, 2020. [286] C. Sturm and W. Wiesbeck, “Waveform design and signal processing [310] X. Cao, J. Xu, and R. Zhang, “Mobile edge computing for cellular- aspects for fusion of wireless communications and radar sensing,” Proc. connected UAV: Computation offloading and trajectory optimization,” IEEE, vol. 99, no. 7, pp. 1236–1259, Jul. 2011. in Proc. IEEE SPAWC, Jun. 2018, pp. 1–5. 33

[311] S. Jeong, O. Simeone, and J. Kang, “Mobile edge computing via [333] B. Nazer and M. Gastpar, “Computation over multiple-access chan- a UAV-mounted cloudlet: Optimization of bit allocation and path nels,” IEEE Trans. Inf. Theory, vol. 53, no. 10, pp. 3498–3516, Oct. planning,” IEEE Trans. Veh. Technol., vol. 67, no. 3, pp. 2049–2063, 2007. Mar. 2018. [334] X. Cao, G. Zhu, J. Xu, and K. Huang, “Cooperative interference [312] Z. Yang, C. Pan, K. Wang, and M. Shikh-Bahaei, “Energy efficient management for over-the-air computation networks,” arXiv preprint resource allocation in UAV-enabled mobile edge computing networks,” arXiv:2007.11765, 2020. IEEE Trans. Wireless Commun., vol. 18, no. 9, pp. 4576–4589, Sep. [335] G. Zhu, J. Xu, and K. Huang, “Over-the-air computing for 6G–turning 2019. air into a computer,” arXiv preprint arXiv:2009.02181, 2020. [313] H. Guo and J. Liu, “UAV-enhanced intelligent offloading for internet of [336] G. Zhu, Y. Wang, and K. Huang, “Broadband analog aggregation for things at the edge,” IEEE Trans. Industrial Informatics, vol. 16, no. 4, low-latency federated edge learning,” IEEE Trans. Wireless Commun., pp. 2737–2746, Apr. 2020. vol. 19, no. 1, pp. 491–506, Jan. 2020. [314] X. Hu, K. Wong, K. Yang, and Z. Zheng, “UAV-assisted relaying and [337] K. Yang, T. Jiang, Y. Shi, and Z. Ding, “Federated learning via over- edge computing: Scheduling and trajectory optimization,” IEEE Trans. the-air computation,” IEEE Trans. Wireless Commun., vol. 19, no. 3, Wireless Commun., vol. 18, no. 10, pp. 4738–4752, Oct. 2019. pp. 2022–2035, Mar. 2020. [315] F. Zhou, Y. Wu, R. Q. Hu, and Y. Qian, “Computation rate max- imization in UAV-enabled wireless-powered mobile-edge computing systems,” IEEE J. Sel. Areas Commun., vol. 36, no. 9, pp. 1927–1941, Sep. 2018. [316] Y. Liu, K. Xiong, Q. Ni, P. Fan, and K. B. Letaief, “UAV-assisted wireless powered cooperative mobile edge computing: Joint offloading, CPU control, and trajectory optimization,” IEEE Internet Things J., vol. 7, no. 4, pp. 2777–2790, Aug. 2020. [317] J. Zhang, L. Zhou, F. Zhou, B. Seet, H. Zhang, Z. Cai, and J. Wei, “Computation-efficient offloading and trajectory scheduling for multi- UAV assisted mobile edge computing,” IEEE Trans. Veh. Technol., vol. 69, no. 2, pp. 2114–2125, Feb. 2020. [318] A. Asheralieva and D. Niyato, “Hierarchical game-theoretic and reinforcement learning framework for computational offloading in UAV-enabled mobile edge computing networks with multiple service providers,” IEEE Internet Things J., vol. 6, no. 5, pp. 8753–8769, Oct. 2019. [319] T. Zeng, O. Semiari, M. Mozaffari, M. Chen, W. Saad, and M. Bennis, “Federated learning in the sky: Joint power allocation and scheduling with UAV swarms,” arXiv preprint arXiv:2002.08196, 2020. [320] Y. Liu, J. Nie, X. Li, S. H. Ahmed, W. Y. B. Lim, and C. Miao, “Federated learning in the sky: Aerial-ground air quality sensing framework with UAV swarms,” IEEE Internet Things J., Early Access. [321] C. Dong, Y. Shen, Y. Qu, Q. Wu, F. Wu, and G. Chen, “UAVs as a service: Boosting edge intelligence for air-ground integrated networks,” arXiv preprint arXiv:2003.10737, 2020. [322] J. S. Ng, W. Y. B. Lim, H.-N. Dai, Z. Xiong, J. Huang, D. Niyato, Qingqing Wu (S’13-M’16) received the B.Eng. and X.-S. Hua, C. Leung, and C. Miao, “Joint auction-coalition formation the Ph.D. degrees in Electronic Engineering from framework for communication-efficient federated learning in UAV- South China University of Technology and Shanghai enabled internet of vehicles,” arXiv preprint arXiv:2007.06378, 2020. Jiao Tong University (SJTU) in 2012 and 2016, [323] O. Esrafilian, R. Gangula, and D. Gesbert, “Learning to communicate in respectively. He is currently an assistant professor UAV-aided wireless networks: Map-based approaches,” IEEE Internet with the State key laboratory of Internet of Things Things J., vol. 6, no. 2, pp. 1791–1802, Apr. 2019. for Smart City, University of Macau. From 2016 to [324] X. Cao, F. Wang, J. Xu, R. Zhang, and S. Cui, “Joint computation 2020, he was a Research Fellow in the Department and communication cooperation for energy-efficient mobile edge com- of Electrical and Computer Engineering at National puting,” IEEE Internet Things J., vol. 6, no. 3, pp. 4188–4200, Jun. University of Singapore. His current research in- 2019. terest includes intelligent reflecting surface (IRS), [325] T. Bai, J. Wang, Y. Ren, and L. Hanzo, “Energy-efficient computation unmanned aerial vehicle (UAV) communications, and MIMO transceiver offloading for secure UAV-edge-computing systems,” IEEE Trans. Veh. design. He was listed as a World’s Top 2% Scientist by Stanford University Technol., vol. 68, no. 6, pp. 6074–6087, Jun. 2019. in 2020. [326] Q. Zhang, J. Chen, L. Ji, Z. Feng, Z. Han, and Z. Chen, “Response He was the recipient of the IEEE Communications Society Young Author delay optimization in mobile edge computing enabled UAV swarm,” Best Paper Award in 2021, the Outstanding Ph.D. Thesis Award of China IEEE Trans. Veh. Technol., vol. 69, no. 3, pp. 3280–3295, Mar. 2020. Institute of Communications in 2017, the Outstanding Ph.D. Thesis Funding in SJTU in 2016, and the IEEE WCSP Best Paper Award in 2015. He was [327] W. Chen, B. Liu, H. Huang, S. Guo, and Z. Zheng, “When UAV swarm the Exemplary Editor of IEEE Communications Letters in 2019 and the meets edge-cloud computing: The QoS perspective,” IEEE Network, Exemplary Reviewer of several IEEE journals. He serves as an Associate vol. 33, no. 2, pp. 36–43, Mar. 2019. Editor for IEEE Communications Letters, IEEE Wireless Communications [328] J. Konecny, H. B. McMahan, F. X. Yu, P. Richtarik, A. T. Suresh, and Letters, IEEE Open Journal of Communications Society (OJ-COMS), and D. Bacon, “Federated learning: Strategies for improving communica- IEEE Open Journal of Vehicular Technology (OJVT). He is the Lead Guest tion efficiency,” arXiv preprint arXiv:1610.05492. Editor for IEEE Journal on Selected Areas in Communications on UAV [329] A. K. Bouziane Brik and M. Bouaziz, “Federated learning for UAVs- Communications in 5G and Beyond Networks, and the Guest Editor for IEEE enabled wireless networks: Use cases, challenges, and open problems,” OJVT on 6G Intelligent Communications and IEEE OJ-COMS on Reconfig- IEEE Access, vol. 8, pp. 53 841–53 849, Mar. 2020. urable Intelligent Surface-Based Communications for 6G Wireless Networks. [330] W. Zhang, L. Li, N. Zhang, T. Han, and S. Wang, “Air-ground He is the workshop co-chair for IEEE ICC 2019-2021 workshop on Integrating integrated mobile edge networks: A survey,” IEEE Access, vol. 8, pp. UAVs into 5G and Beyond, and the workshop co-chair for IEEE GLOBECOM 125 998–126 018, 2020. 2020 and ICC 2021 workshop on Reconfigurable Intelligent Surfaces for [331] X. Cao, G. Zhu, J. Xu, and K. Huang, “Optimized power control for Wireless Communication for Beyond 5G. He serves as the Workshops and over-the-air computation in fading channels,” IEEE Trans. Wireless Symposia Officer of Reconfigurable Intelligent Surfaces Emerging Technology Commun., Early Access. Initiative and Research Blog Officer of Aerial Communications Emerging [332] G. Zhu and K. Huang, “Mimo over-the-air computation for high- Technology Initiative. He is the IEEE ComSoc Young Professional Chair in mobility multimodal sensing,” IEEE Internet Things J., vol. 6, no. 4, APB. pp. 6089–6103, Aug. 2019. 34

Jie Xu (S’12-M’13) received the B.E. and Ph.D. Derrick Wing Kwan Ng (S’06-M’12-SM’17-F’21) degrees from the University of Science and Technol- received the bachelor degree with first-class honors ogy of China in 2007 and 2012, respectively. From and the Master of Philosophy (M.Phil.) degree in 2012 to 2014, he was a Research Fellow with the electronic engineering from the Hong Kong Uni- Department of Electrical and Computer Engineering, versity of Science and Technology (HKUST) in National University of Singapore. From 2015 to 2006 and 2008, respectively. He received his Ph.D. 2016, he was a Post-Doctoral Research Fellow with degree from the University of British Columbia the Engineering Systems and Design Pillar, Singa- (UBC) in 2012. He was a senior postdoctoral fel- pore University of Technology and Design. From low at the Institute for Digital Communications, 2016 to 2019, he was a Professor with the School of Friedrich-Alexander-University Erlangen-Nurnberg¨ Information Engineering, Guangdong University of (FAU), Germany. He is now working as a Senior Technology, China. He is currently an Associate Professor with the School of Lecturer and a Scientia Fellow at the University of New South Wales, Sydney, Science and Engineering, The Chinese University of Hong Kong, Shenzhen, Australia. His research interests include convex and non-convex optimization, China. His research interests include wireless communications, wireless physical layer security, IRS-assisted communication, UAV-assisted communi- information and power transfer, UAV communications, edge computing and cation, wireless information and power transfer, and green (energy-efficient) intelligence, and integrated sensing and communication (ISAC). He was a wireless communications. recipient of the 2017 IEEE Signal Processing Society Young Author Best Dr. Ng received the Australian Research Council (ARC) Discovery Early Paper Award, the IEEE/CIC ICCC 2019 Best Paper Award, the 2019 IEEE Career Researcher Award 2017, the Best Paper Awards at the WCSP 2020, Communications Society Asia-Pacific Outstanding Young Researcher Award, IEEE TCGCC Best Journal Paper Award 2018, INISCOM 2018, IEEE and the 2019 Wireless Communications Technical Committee Outstanding International Conference on Communications (ICC) 2018, IEEE International Young Researcher Award. He is the Symposium Co-Chair of the IEEE Conference on Computing, Networking and Communications (ICNC) 2016, GLOBECOM 2019 Wireless Communications Symposium, the workshop co- IEEE Wireless Communications and Networking Conference (WCNC) 2012, chair of several IEEE ICC and GLOBECOM workshops, the Tutorial Co- the IEEE Global Telecommunication Conference (Globecom) 2011, and the Chair of the IEEE/CIC ICCC 2019, and the Founding Chair of the IEEE IEEE Third International Conference on Communications and Networking in WTC Special Interest Group (SIG) on ISAC. He served or is serving as China 2008. He has been serving as an editorial assistant to the Editor-in-Chief an Editor of the IEEE TRANSACTIONS ON COMMUNICATIONS, IEEE of the IEEE Transactions on Communications from Jan. 2012 to Dec. 2019. WIRELESS COMMUNICATIONS LETTERS, and Journal of Communica- He is now serving as an editor for the IEEE Transactions on Communications, tions and Information Networks, an Associate Editor of IEEE ACCESS, the IEEE Transactions on Wireless Communications, and an area editor for and a Guest Editor of the IEEE WIRELESS COMMUNICATIONS, IEEE the IEEE Open Journal of the Communications Society. Also, he has been JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, and Science listed as a Highly Cited Researcher by Clarivate Analytics since 2018. China Information Sciences.

Yong Zeng (S’12-M’14) is with the National Mo- bile Communications Research Laboratory, South- east University, China, and also with the Purple Mountain Laboratories, Nanjing, China. He received the Bachelor of Engineering (First-Class Honours) and Ph.D. degrees from Nanyang Technological Uni- versity, Singapore, in 2009 and 2014, respectively. From 2013 to 2018, he was a Research Fellow and Senior Research Fellow at the Department of Electri- cal and Computer Engineering, National University of Singapore. From 2018 to 2019, he was a Lecturer at the School of Electrical and Information Engineering, the University of Sydney, Australia. Dr. Zeng was listed as 2020 and 2019 Highly Cited Researcher by Clarivate Naofal Al-Dhahir is Erik Jonsson Distinguished Analytics. He is the recipient of the Australia Research Council (ARC) Professor & ECE Dept. Associate Head at UT- Discovery Early Career Researcher Award (DECRA), 2020 IEEE Marconi Dallas. He earned his PhD degree from Stanford Prize Paper Award in Wireless Communications, 2018 IEEE Communications University and was a principal member of technical Society Asia-Pacific Outstanding Young Researcher Award, 2020 & 2017 staff at GE Research Center and AT&T Shannon IEEE Communications Society Heinrich Hertz Prize Paper Award. He serves Laboratory from 1994 to 2003. He is co-inventor as an Associated Editor for IEEE Communications Letters and IEEE Open of 43 issued patents, co-author of about 470 pa- Journal of Vehicular Technology, Leading Guest Editor for IEEE Wireless pers and co-recipient of 4 IEEE best paper awards. Communications on “Integrating UAVs into 5G and Beyond” and China He is an IEEE Fellow, received 2019 IEEE SPCC Communications on “Network-Connected UAV Communications”. He is the technical recognition award and 2021 Qualcomm workshop co-chair for ICC 2018-2021 workshop on UAV communications, faculty award. He served as Editor-in-Chief of IEEE the tutorial speaker for Globecom 2018/2019 and ICC 2019 tutorials on UAV Transactions on Communications from Jan. 2016 to Dec. 2019. He is a Fellow communications. of the National Academy of Inventors. 35

Robert Schober (S’98-M’01-SM’08-F’10) received the Diplom (Univ.) and the Ph.D. degrees in electri- cal engineering from Friedrich-Alexander University of Erlangen-Nuremberg (FAU), Germany, in 1997 and 2000, respectively. From 2002 to 2011, he was a Professor and Canada Research Chair at the University of British Columbia (UBC), Vancouver, Canada. Since January 2012 he is an Alexander von Humboldt Professor and the Chair for Digital Communication at FAU. His research interests fall into the broad areas of Communication Theory, Wireless Communications, and Statistical Signal Processing. Robert received several awards for his work including the 2002 Heinz Maier Leibnitz Award of the German Science Foundation (DFG), the 2004 Innovations Award of the Vodafone Foundation for Research in Mobile Communications, a 2006 UBC Killam Research Prize, a 2007 Wilhelm Friedrich Bessel Research Award of the Alexander von Humboldt Foundation, the 2008 Charles McDowell Award for Excellence in Research from UBC, a 2011 Alexander von Humboldt Professorship, a 2012 NSERC E.W.R. Stacie Fellowship, and a 2017 Wireless Communications Recognition Award by the IEEE Wireless Communications Technical Committee. Since 2017, he has been listed as a Highly Cited Researcher by the Web of Science. Robert is a Fellow of the Canadian Academy of Engineering, a Fellow of the Engineering Institute of Canada, and a Member of the German National Academy of Science and Engineering. From 2012 to 2015, he served as Editor-in-Chief of the IEEE Transactions on Communications. Currently, he serves as Member of the Editorial Board of the Proceedings of the IEEE and as VP Publications for the IEEE Communication Society (ComSoc).

A. Lee Swindlehurst received the B.S. (1985) and M.S. (1986) degrees in Electrical Engineering from Brigham Young University (BYU), and the PhD (1991) degree in Electrical Engineering from Stanford University. He was with the Department of Electrical and Computer Engineering at BYU from 1990-2007, where he served as Department Chair from 2003-06. During 1996-97, he held a joint appointment as a visiting scholar at Uppsala University and the Royal Institute of Technology in Sweden. From 2006-07, he was on leave working as Vice President of Research for ArrayComm LLC in San Jose, California. Since 2007 he has been a Professor in the Electrical Engineering and Computer Science Department at the University of California Irvine, where he served as Associate Dean for Research and Graduate Studies in the Samueli School of Engineering from 2013-16. During 2014-17 he was also a Hans Fischer Senior Fellow in the Institute for Advanced Studies at the Technical University of Munich. In 2016, he was elected as a Foreign Member of the Royal Swedish Academy of Engineering Sciences (IVA). His research focuses on array signal processing for radar, wireless communications, and biomedical applications, and he has over 300 publications in these areas. Dr. Swindlehurst is a Fellow of the IEEE and was the inaugural Editor-in-Chief of the IEEE Journal of Selected Topics in Signal Processing. He received the 2000 IEEE W. R. G. Baker Prize Paper Award, the 2006 IEEE Communications Society Stephen O. Rice Prize in the Field of Communication Theory, the 2006 and 2010 IEEE Signal Processing Societys Best Paper Awards, and the 2017 IEEE Signal Processing Society Donald G. Fink Overview Paper Award.